On the third take, in Johannesburg's first recording studio, sometime in 1939, a Zulu migrant worker named Solomon Linda found the notes. Not planned. Not constructed from theory. Found — the way the best things are found, by arriving at a place you didn't know you were traveling toward and recognizing it only after you are already there. The melody floated up in falsetto above the rolling choral drive of his group, the Evening Birds, and landed in the air of that room and stayed there, and has been staying there, in various forms and under various names, in the ears of billions of people who have never heard his name. He was paid a small one-off fee by Gallo Records. The recording rights went to the label. He died in Johannesburg in 1962, in poverty, without knowing that the notes he found on the third take would become the most widely recognized melody ever to emerge from Africa. His daughter Adelaide died in 2001, at thirty-eight years old, because she could not afford antiretroviral medication. The song was being performed, at that moment, in Disney's The Lion King stage production, which had been running on Broadway since 1997 and had by then generated tens of millions of dollars in revenue from a melody that her father found on the third take in 1939.
The question is not legal. The legal question was eventually answered, partially, in a 2006 settlement — confidential terms, a lump sum for past royalties, a share of future revenue, a co-writing credit for Solomon Linda on a song he wrote alone. The question is moral. And the moral question is not whether the system failed. The moral question is whether the system worked exactly as intended — and whether we have the honesty to say so.
He comes back, Linda. Not dramatically. Not in the cinematic version where the ghost appears in the mirror or the lights flicker. He comes back the way all cultural ghosts come back — through the children singing a song they didn't know had a source. Through the classroom that teaches the Disney version of Africa and shows no photograph of the man who made it possible. Through every streaming payment that goes somewhere else while the melody runs. Through every production that invokes his work and does not invoke his name. He is the most present absent person in the history of popular music. The echo outlived the man. The echo always does. The question is who collects the revenue from the echo — and what that answer says about us.
Human Culture Is One Long Act of Sampling — And Everyone Knows It
Here is the thing nobody at the copyright conference wants to say plainly: sampling is not a modern invention. Sampling is not a hip-hop invention. Sampling is not even a music invention. Sampling is the mechanism by which human culture reproduces itself, and it has been operating since the first story was told around the first fire by someone who heard a different story the week before and changed it slightly and called it theirs.
The words you are reading right now were invented by someone else. The metaphors were borrowed from image systems that precede any individual writer's life. The myths that give your culture its shape came from somewhere. The melodies that feel ancient and universal came from specific people in specific rooms who improvised them and passed them forward. Jazz borrowed from blues, which borrowed from field hollers and gospel, which borrowed from West African polyrhythmic traditions that were themselves transformations of older forms stretching back to origins nobody remembers because the origins were never written down because the people holding them were not considered worth writing about.
Greek mythology borrowed from Egyptian cosmology. Christianity borrowed from Judaism, which borrowed from Babylonian creation narratives. The Epic of Gilgamesh contains a flood story that predates Genesis by a thousand years. Shakespeare borrowed plots from Plutarch, from Italian novellas, from English chronicles. He invented the language he wrote them in, in the sense that he bent and expanded what existed until it became something new — but the raw materials were not his. The raw materials were never anyone's. The raw materials are the accumulated residue of every human mind that ever encountered the world and tried to make sense of it in language.
Human culture is a relay race of transformation. Every baton was passed. Every baton was also, at the moment of passing, changed slightly by the person passing it. That is not theft. That is transmission. That is how the thing stays alive across centuries rather than calcifying into a museum piece nobody engages with.
The problem is not the sampling. The problem is who gets paid for it — and who gets erased in the process.
The Ghost Choir — What AI Is Actually Made Of
When an AI generates music in 2026, what exactly is it doing? The technical answer is: calculating statistical probabilities across a training dataset of human creative output, predicting which sonic or linguistic element is most likely to follow any given input based on patterns learned from millions of examples. The honest answer is: sampling everything that has ever been made, simultaneously, without attribution, at a scale that makes the most ambitious hip-hop producer look like someone borrowing a cup of sugar.
The AI's training data is the ghost choir. Part Solomon Linda. Part the Zulu oral traditions that shaped his musicality before he ever stood in front of a microphone. Part the unknown composers of every breakbeat that Clyde Stubblefield played into existence on James Brown's recordings at King Studios in Cincinnati. Part the teenager who posted a voice memo to SoundCloud at 2 AM that forty people heard and that became, in some unmappable way, part of the statistical distribution the model learned from. Part the failed songwriter whose unpublished demos were scraped from a hard drive without consent. Part the grandmother whose lullabies were never recorded but whose granddaughter recorded a cover version that was uploaded and absorbed. Part every tradition, every region, every century, compressed into a probability matrix that outputs melodies when prompted.
If Solomon Linda's original recording of Mbube was included in an AI's training data — and it almost certainly was — then every AI-generated melody that carries the faintest echo of that falsetto line is, in some distributed and untraceable sense, built partly on his work. With no credit. With no compensation. With no mechanism for his family to know it happened. The digital age is not eliminating the extraction problem. It is automating it at planetary scale.
The Extraction Playbook — Three Centuries, Same Structure
The language updates. The mechanism does not.
The Counterargument That Has to Be Taken Seriously
Here is where the essay has to argue with itself, because if it doesn't, it becomes advocacy rather than analysis — and advocacy, however righteous, is less useful than the version that has already absorbed the strongest objection.
The counterargument is this: perhaps originality is real. Not in the sense of creating from nothing — nobody creates from nothing, the universe of available materials is finite and inherited — but in the sense of arranging existing elements through a consciousness that has never existed before and will never exist again. Nobody invented notes. Nobody invented words. Nobody invented pigments. Yet nobody wrote Purple Rain before Prince Rogers Nelson wrote it. Nobody wrote Hamlet before William Shakespeare did. Nobody painted the series of wheat fields Vincent van Gogh completed in the weeks before his death in Auvers-sur-Oise in July 1890, in a state of extraordinary productive intensity that was either mania or genius or both, producing seventy-five paintings in seventy-five days, each one unmistakably his and unmistakably nobody else's.
The notes were borrowed. The genre was inherited. The biographical suffering that charged the work was, in some sense, the universal suffering of being alive and sensitive and unequipped by evolution to process the intensity of existence. And yet what emerged was his. Irreducibly, specifically, non-reproducibly his — even though every element was technically antecedent. Perhaps originality is not creating something from nothing. Perhaps it is arranging inherited things through a consciousness that exists once, in one direction, and then doesn't. And perhaps that non-repeatability is worth protecting, worth compensating, worth crediting — not because the inputs were owned, but because the specific arrangement of them was uniquely possible only through one particular combination of life, suffering, and time.
If that is true, then Solomon Linda's falsetto line — improvised on the third take, emerging from a specific body carrying specific Zulu musical traditions through a specific childhood in a specific colonial context — was original in the deepest sense. Not because the scale was invented. Because that particular configuration of those particular notes on that particular afternoon could only have come from him. And if that is what we mean by originality, then its theft is not merely an economic injustice. It is an ontological one. Something that could only have existed once was taken and its existence was attributed to everyone except the person who made it possible.
The Absurdity That Nobody Laughs About
At some point in this history, someone at a company whose name is familiar to every child on earth sat in a meeting and made a decision about how to handle the Mbube situation, and the decision they reached — for years, through multiple legal challenges, through the knowledge that Linda's daughters were living in poverty while the song ran in a production grossing hundreds of millions of dollars — was to do as little as legally required for as long as legally possible. This is not a story about a cartoon villain in a boardroom. The people in that meeting were almost certainly reasonable, pleasant, family-oriented individuals who attended charity events and remembered birthdays and were, in their private lives, generous in the ways that people with disposable income can afford to be generous. The decision they made was structural, not sadistic. The structure said: protect the asset. Minimize the liability. Pay what the law requires and not a dollar more. The structure was not evil. The structure was a machine running its program. The machine's program was not designed to ask whether Solomon Linda's daughter could afford medication. That question was outside the scope of the program.
The darkly funny thing — the thing that requires a specific quality of rueful laugh that has no clean name — is that the same company has built its entire brand identity around stories about the importance of following your dreams, believing in yourself, and never giving up on what you love. The lion sleeps tonight. The dreamer dreams on. The grandfather's melody soundtracks the grandson's cinema. The irony is structural rather than intentional, which makes it worse, not better. Intentional irony has an author. Structural irony just runs.
He comes back again. He has to. Because his daughters signed away the rights twice — once in 1983, once in 1992, each time for approximately one dollar, to a company that knew exactly what those rights were worth and understood exactly what it was acquiring. The legal documents exist. They are a matter of public record. Someone prepared those documents. Someone presented them to women who had no lawyers and no understanding of US copyright law and were living in poverty in South Africa. Someone watched them sign. The documents were filed. The machine kept running. Adelaide died in 2001 at thirty-eight. The Lion King stage production ran to standing ovations in more than a hundred cities. The melody continued.
What the Law Was Actually Designed For
Copyright law is frequently discussed as if it were designed to protect creators. This requires a selective reading of the history. Early intellectual property frameworks in the eighteenth and nineteenth centuries were designed primarily to protect publishers and distributors — the people who controlled the infrastructure of reproduction, who had capital at risk in the printing or pressing process, who needed legal exclusivity to justify that capital investment. The artist's interest was, at best, secondary. At worst, the artist's interest was the legal pretext for a framework that primarily served the interests of whoever held the distribution infrastructure.
What this produced, structurally, was a system in which the person who made the thing and the person who owned the rights to the thing were frequently different people — and in which the mechanisms for transferring those rights were optimized for speed and simplicity rather than for informed consent by parties with equivalent bargaining power. Linda signed away his rights in 1939 in apartheid South Africa, where Black artists had no negotiating position, no legal representation, and no context for understanding what international copyright might mean for a recording that would be heard on every continent within thirty years. His wife and daughters signed them away again in 1983 and 1992, for one dollar each time, to a company that had the full benefit of legal counsel and decades of experience in exactly this kind of transaction. This was not a failure of the legal system. This was the legal system operating as designed, allocating resources to the parties best positioned to exploit them, which is what market systems do when they are not constrained by other considerations.
The question Socrates would have asked — not what is legal, but what is just — is one that the legal system was not built to answer. Justice is not within the system's scope. The system's scope is enforcing the terms of agreements. Whether the agreements were made between parties with equivalent power and information is a different question, and the answer to that different question is, in the case of Solomon Linda, and in the case of most Black creative artists who encountered the music industry during the twentieth century, no.
Was Originality Ever Real? — The Question That Ruins the Party
If we accept the argument that human culture is a relay race of transformation — that every song contains other songs, every story contains other stories, every mind contains other minds — then the question of who owns any given creative work becomes genuinely unstable. Not legally unstable. Philosophically unstable. If Mbube contains Zulu choral tradition, which contains ancestral practices going back centuries, which contain influences from neighboring cultures, which contain — at some untraceable point — the same human impulse toward communal music-making that produced Gregorian chant in Europe and court music in Tang Dynasty China and ceremonial drums in the Kingdom of Dahomey simultaneously, then who exactly has the standing to own a particular crystallization of that impulse?
This is the question that makes the whole courtroom uncomfortable, because it suggests that every intellectual property claim is, at some level, a claim made by a specific historical moment against an infinite continuum, and the claim is always somewhat arbitrary in its choice of where to draw the line. We draw the line at the individual creator. We say: this person, at this moment, arranged these elements in this specific way, and that specific arrangement belongs to them. That is a reasonable place to draw the line. It is not the only place, and it is not a place that has any claim to being derived from natural law rather than from a series of historical decisions made by people with particular interests.
What the Linda case does — what makes it more than a story of one man's misfortune — is expose how that line-drawing functions in practice when the power differential between the parties is extreme. The line gets drawn wherever the party with the most legal resources is standing. That is the honest version of the history.
What a Different System Might Look Like
The interesting thing about the current moment — the uncomfortable thing, the thing that makes it worth talking about rather than simply lamenting — is that the technology now exists to make attribution and compensation scalable in ways they were not in 1939, or 1983, or even 2004 when the Linda family filed their lawsuit. Blockchain-based rights registries can record the chain of derivation for any creative work, allowing royalties to flow not just to the most recent owner of a copyright but to every contributor in the lineage of a work's development. AI training data can be catalogued and attributed in ways that create legal and financial obligations to the creators whose work contributed to the model's capabilities. Dynamic licensing systems can replace the binary of total ownership or total public domain with graduated participation models that compensate original creators while allowing transformation and remixing to continue.
None of this happens automatically. None of it happens because the technology enables it. It happens only if the people with the power to implement it decide that the current extraction model produces outcomes they are no longer comfortable defending — and that moment may or may not arrive before the current model has extracted everything available from the current generation of uncredited sources.
The Echo
Every time a child sings "The Lion Sleeps Tonight" — in a classroom, in a car, in the shower, in the particular unselfconscious way that children sing songs they love without knowing anything about the history of those songs — Solomon Linda breathes again, briefly. The melody escapes whatever legal and economic structure has been built around it and does the thing that melodies do, which is travel through human nervous systems without asking permission and lodge in the memory and become part of the internal soundtrack of a life. That cannot be owned. That has never been owned. The experience of hearing music and being changed by it, however slightly, is not a licensed transaction. It is something older and stranger and more interesting than anything the copyright system was designed to regulate.
But the money can be owned. The credit can be controlled. The historical record can be curated. A man can become a footnote to his own creation — can be so thoroughly erased from the story of what he made that billions of people carry his melody in their bodies without knowing his name — and this can happen not through any single act of malice but through the ordinary, impersonal, systemic operation of a machine designed to move value in a particular direction.
The ghost choir of human creative output — the vast, layered, centuries-deep accumulation of everything anyone ever made and transmitted — is now being processed by systems that can generate new combinations from it faster than any human can and without any mechanism for attributing the inheritance. That is where we are. Solomon Linda's voice, absorbed into a training dataset, producing synthetic melodies carrying faint traces of a falsetto improvised on the third take in Johannesburg in 1939, with no credit and no payment to the family that is still, in 2026, navigating the aftermath of agreements made for one dollar in the 1980s.
The question is not whether the ghost choir is singing. It obviously is. The question is whether we intend to keep pretending we can't hear it.
And whether we can look at what happened to Solomon Linda — at the full, specific, documented record of it, not the sanitized version, the one with Adelaide in it, dying at thirty-eight in 2001 while the Broadway lights were on — and call the system that produced that outcome one we are willing to defend.
The melody is still going.
He found it on the third take.
It belongs, in some sense that no court has ever been able to fully settle, to everyone and no one.
It belongs, in the sense that should matter most, to him.