In 1713, Johann Sebastian Bach sat down at his desk in Weimar and began copying out concertos by Antonio Vivaldi. He transcribed them note for note, in his own hand, working through at least nine of the L'estro armonico concertos like a medical student dissecting a cadaver. The work was painstaking, derivative on its face, and (as it turned out) the foundation of everything Bach would become. Once he'd absorbed the architecture of the Italian concerto, he produced the Brandenburg Concertos, music that sounds nothing like Vivaldi and could only have come from Bach.

The conventional warning is that you shouldn't copy because copying is theft. Austin Kleon's bestseller (Steal Like an Artist, 2012) tried to rescue copying from this stigma by reframing it as the basis of all creative work. He was right; but what does the heist look like, when it works?

The transcription test

When you copy a Vivaldi concerto into a manuscript by hand, you're not producing a Vivaldi concerto - nobody would think that. But you are forced to interpret a thousand small decisions about how the music coheres, why one voice enters on a particular beat, what makes the ritornello structure hold. You learn, through your fingers, why something works.

Hunter S. Thompson did this with prose. In his early twenties, working as a copyboy at Time magazine, he typed out the complete texts of The Great Gatsby and A Farewell to Arms. The point was never to plagiarise Fitzgerald or Hemingway; Thompson wanted to feel, in his hands, the weight of a good sentence.

Two motives separate good copying from bad copying.

One copies to produce a finished thing. The other copies to understand a process. The first motive yields imitations; the second yields apprenticeships.

What surface imitation gets wrong

A young writer who reads Cormac McCarthy and decides the trick is zero quotation marks is making a classic error. The aspiring filmmaker who thinks Wes Anderson reduces to symmetrical framing and warm yellow tones is doing the same. Steve Jobs gets reverse-engineered into a black turtleneck; Brian Eno collapses into longer reverb tails. Etc.

These are cargo cult creators.

The South Pacific islanders who built bamboo control towers after the war ended believed that if they replicated the surface details of an American airstrip, the planes would return with cargo. The towers were elaborate, and the construction careful. But the planes never come back.

Cargo cult creativity makes the same error. It assumes the visible artefact is the cause of the result, when the visible artefact is the consequence of something deeper and hidden. McCarthy's omitted punctuation is a consequence of how he thinks about voice, about the unbroken pressure of a sentence, about what a comma costs in narrative momentum. Strip the punctuation from another writer's prose and you don't get McCarthy; you get a manuscript that's harder to read for no good reason.

The imitators fail because they copy the wrong layer.

Influence as collision

Picasso said he wanted to draw like Raphael, and then spent his life learning to draw like a child. In 1957, at age 75, he locked himself in his villa in Cannes and produced 58 paintings reinterpreting Velázquez's Las Meninas (1656). He preserved the composition and the spatial relationships; the cast remained intact: the infanta, the dwarf, the dog, and the painter in the background. He produced a body of work that no one would mistake for Velázquez.

The Velázquez composition served as the immovable object. Picasso's cubist habits and Spanish political grief served as the moving force. The 58 paintings record what happened at the point of impact.

A violent collision is closer to the actual mechanism of influence than any of the polite formulations about being "inspired by" something. Influence operates as collision; the thing you make becomes original to the degree that your own concerns and limitations distort whatever you took in. Bowie cut up his lyrics using William Burroughs's method. Hilary Mantel built Wolf Hall on the bones of Holinshed's Chronicles. Sondheim wrote Sweeney Todd through the lens of Bernard Herrmann's film scores. Tarantino made an entire career out of the friction between exploitation cinema and an obsessive's encyclopaedic memory of it.

Their motive was to absorb something specific and put it under pressure. The originality came out the other end as a by-product.

James Joyce's Ulysses (1922) maps, episode by episode, onto Homer's Odyssey, written some twenty-seven centuries earlier. Joyce kept the scaffolding intact: the wanderer, the long way home, the underworld, the encounter with the dead, the suitors at the gate. He even prepared a private schema (the Linati and Gilbert schemas) mapping each chapter to a Homeric episode, an organ of the body, a colour, and a technique. The borrowing was total. And the result is a book that nobody mistakes for Homer, because the friction between an ancient epic and a single day in 1904 Dublin produced something the original Greek could never have generated. The structural borrowing freed Joyce to do what was actually new: a prose technique that bends to fit each chapter's subject. He took the bones precisely so he could invent the flesh.

The two questions

When you encounter a piece of work that moves you, there's a fork in the road. You can ask "what does this look like?" or you can ask "why does this work?"

Saul Bellow once said that a writer is a reader moved to emulation. The emulation he meant was the urge to do, yourself, what you had felt done to you by a great book. That urge has very little to do with sentence structure and a lot to do with effect: I want my readers to feel this thing I just felt.

The means come second.

If you can articulate what a piece of work is trying to do, you can borrow its method without inheriting its appearance. But if you can describe only what the work looks like, you'll produce a version that looks awfully similar and actually does nothing.

The fingerprint problem

Beginners worry about whether their work is original. Pros worry about whether their work is honest. These anxieties are incompatible - and they produce different working lives.

Originality, as a goal, is incoherent. Nobody starts from nothing. Every painter has seen paintings. Every writer has read books. The mind that produces "original" work is a mind that's already been shaped by thousands of inputs, most of which it can no longer name.

What we recognise as a personal style is the particular ratio of influences. Murakami plus Carver plus jazz plus marathon running is what produces a Murakami. Subtract any of those and you get someone else.

The question "is my work too derivative?" tends to be the wrong question entirely. The better question is whether you've absorbed enough different things, and absorbed them with enough seriousness, that the combination is uncopyable. A writer who's only ever read three contemporary literary novelists will produce derivative work no matter how hard he pushes against it. A writer who's read those three plus Cicero plus Joan Didion plus theological treatises plus aviation manuals has too many vectors to ever come out sounding like any one of them.

Range is the cure.

When the surface is free

For most of history, surface-level imitation took a good deal of effort. Only a talented painter in their right could forge the work of another - producing something that looked like a Caravaggio meant spending years learning to paint. Producing something that read like Hemingway meant writing thousands of pages of bad sentences first. The skill needed to imitate the surface was, in itself, the apprenticeship that taught you the depth. Surface and substance were welded together by the cost of the work.

Large language models have broken that arrangement; a model can produce a thousand-word piece in the style of any well-known writer, with the correct surface features, in about four seconds. The barrier to imitation is crumbling.

What then?

When anyone can summon the surface of any voice in seconds, the surface becomes a commodity; what remains scarce is the underlying judgement - what to say in that voice, what to leave out, why a particular reference earns its place, when to drop the voice entirely.

The visible parts are now cheap.

The invisible parts are the whole ballgame.

What the apprenticeship actually involves

There is no shortcut. Not to this, not to anything.

If you want to be inspired without copying, you have to spend time inside the work that moves you. Skimming it doesn't count, and neither does gesturing at it on a podcast. You have to sit with it long enough to map its decisions, identify its constraints, and understand what got rejected as well as much as what got kept. You have to be able to articulate the work in your own terms before you can transmute it into your own forms.

The writer who reads only writers, the designer who looks only at design, the founder who studies only founders, is starving the engine. The collisions that produce something new tend to come from oblique angles. The richest period of European painting (the Dutch Golden Age) overlapped with a flood of trade in spices, optics, cartography, and lens-making. Vermeer, after all, learned about light from instrument-makers; originality is downstream of variety. It must be.

You have to be patient with the gap. Years of input precede any output worth keeping. Bach copied Vivaldi for years before the Brandenburgs. Picasso painted in classical mode for two decades before cubism. Joni Mitchell played other people's standards in coffee houses for years before Blue. Hunter Thompson typed out The Great Gatsby and then spent fifteen years writing journalism that no one could ever mistake for Fitzgerald.

These are long, long stretches of work that looked, from outside, like nothing was happening. Inside, the inputs were being broken down into their components, sorted, and rebuilt as something the practitioner could call their own. The temptation, especially now, is to skip this phase by trusting a model to deliver the surface without the years. That temptation should be refused for the same reason a virtuoso refuses to lip-sync: the work that bypasses the apprenticeship produces no apprentice, only an output. And an output is not enough.

The work you're trying to do hasn't been done yet, because you haven't done it. Nobody else will do it for you. The route is through, not around.

In a few hundred years, someone might transcribe your work by hand to understand how it holds together. That's not a bad goal, in itself.