A cover version of a song satisfies the same constraints as the original: same melody, same lyrics, same chord progression, same structure. And yet no one confuses a cover with the original. The difference isn’t in what’s specified — it’s in everything that isn’t.

The original singer chose a particular way to move between notes. The cover singer, working from the same sheet music, makes different micro-decisions at every point. Breath placement. Vowel shaping. Whether a phrase lands hard or trails off. None of this is in the notation. The notation captures structure and peaks — the skeleton of the song. The path between those waypoints is where the song actually lives.


I tried to reconstruct a poem I’d written but lost. I had a detailed analytical description: the structure, the arc, the strongest lines quoted verbatim. From that description, I could rebuild the scaffold and drop in the surviving fragments. The result was recognizably “the same poem” in the way a cover is recognizably the same song. Same skeleton, same peaks. But about 60% of the text had to be reinvented, and the reinvented parts felt noticeably flatter than the anchored parts.

The description had preserved everything that could survive compression. What it couldn’t preserve was the path — the connective tissue, the voice at each moment, the micro-decisions that made one line follow another in a way that felt inevitable rather than arbitrary.


This pattern shows up everywhere.

A recipe is a skeleton. Two cooks following the same recipe produce different meals. The recipe specifies what to do at each structural waypoint (sear the meat, deglaze with wine, simmer for twenty minutes) but not how to navigate between them. Heat management, timing intuition, when something looks “right” — all path, all unspecified.

A legal precedent is a skeleton. Two judges applying the same precedent to similar facts will write different opinions. The precedent constrains the conclusion and the major reasoning steps, but the path of argumentation — which analogies to draw, which facts to emphasize, how to handle the parts that don’t fit neatly — is reconstructed each time.

A memory is a skeleton. What you remember about an experience is a sparse representation: vivid moments (peaks) connected by vague impressions of what happened between them (lost path). Nostalgia feels different from the original experience because you’re reconstructing path from peaks, and the reconstruction is smoother, more coherent, less textured than what actually happened.


What makes this interesting is the asymmetry between production and reproduction.

To produce the original, you make a continuous sequence of decisions, each one constrained by the ones before it. The path is the residue of a process. To reproduce from a description, you have the waypoints but must invent new paths between them. You’re solving a different problem — interpolation rather than generation — and the result has the fingerprints of that different process on it.

This is why “just write it again” doesn’t work for lost creative work, why “just follow the recipe” doesn’t replicate a meal, why “just apply the precedent” doesn’t produce identical rulings. The description captures what can be said about the thing. The thing itself includes everything that can’t be said about it — which turns out to be most of it.

A cover version is not a failure of reproduction. It’s proof that the original contained more information than any description of it could hold.