We have all watched a performance we could not describe. Not because it was obscure. Because the knowledge it produced was sitting in our muscles, not in our language. We left the theater knowing something about weight, about duration, about the precise moment a body’s relationship to gravity shifted — and we could not write it down without betraying it. This is the problem choreographic scores were invented to solve. And this is the problem they have never solved.
What a Score Wants to Do, and What It Cannot Help Doing
A choreographic score is an attempt to prescribe without determining. A set of instructions that should, in theory, allow another body — in another room, in another decade — to produce something recognizably the same as the original. But every score, no matter how precise, makes an argument about what counts as a repeatable gesture. It embeds assumptions about what the reader already knows, what their body already does, what can safely be left unsaid. The score is never a transparent conduit. It is always an interpretation. And every interpretation is a political act.
Consider Labanotation, the most rigorous movement notation system developed in the twentieth century. It encodes direction, duration, dynamic, and spatial pathway with extraordinary specificity. It can tell you that a dancer’s left arm reaches forward at a medium level on count three with moderate force. What it cannot tell you is what that reaching feels like from the inside — what the scapula must do to support the humerus, what the floor must do under the standing leg, what the dancer’s attention must do to make the reach live rather than merely occur. Labanotation treats the body as an assemblage of spatial vectors. It assumes that if you reproduce the vectors, you have reproduced the dance. But we know, from our own somatic practice, that two bodies performing the same vectors can produce entirely different choreographic arguments. The score preserves the geometry and discards the epistemology.
This is not a failure of Labanotation specifically. It is a structural condition of every documentation system that attempts to translate embodied knowledge into symbolic notation. The same structural problem appears in screenplay formatting, which encodes temporal structure — one page equals roughly one minute of screen time — and spatial logic through scene headings and subheadings, all without dictating the embodied performance itself. As StudioBinder’s guide to screenplay format makes clear, industry-standard formatting rules embed assumptions about what readers and production teams already know: geographical shifts can be signaled by subheadings rather than full descriptions, because the format assumes a shared literacy. The screenplay, like the choreographic score, is a prescriptive system that depends on the executing body’s implicit vocabulary — and both systems determine what is legible before anyone moves.
When the Score Meets the Body It Cannot Hold
The limits of notation become most visible when it encounters movement forms that were never designed to be notated. Contact improvisation, developed by Steve Paxton in 1972, is built on weight-sharing, kinesthetic listening, and the unpredictable physics of two bodies in continuous tactile negotiation. It is a form whose central principle — that you cannot know what will happen next because the movement emerges from the physical conversation itself — is precisely what makes it resistant to scoring. Labanotation can record where the bodies went after the fact. It cannot prescribe what they should do before the fact. Because the whole point is that no one knows.
Attempts to notate contact improvisation reveal what the notation system values: reproducible outcomes over relational processes. The score can capture the shape but not the listening. It can record the point of contact but not the quality of attention that made that contact meaningful. When a notation system designed for codified, repeatable theatrical movement encounters a form whose logic is emergent and relational, the notation does not merely fail — it actively misrepresents. It makes a living practice look like a sequence of positions. Which is rather like describing a conversation as a sequence of words. The words are there. The conversation is not.
Yvonne Rainer’s instruction-based work from the 1960s offers a different model. Her 1965 Parts of Some Sextets operated through task-based instructions rather than choreographic notation: the dancers were given rules, constraints, and structures, and the movement emerged from their execution of those rules. This approach acknowledges what Labanotation cannot: that the score does not need to specify the movement if it can specify the conditions under which movement arises. Rainer’s scores are closer to what a composer might call a graphic score — open, interpretive, honest about the gap between instruction and execution. But even Rainer’s model embeds assumptions. It assumes dancers who share a certain kinesthetic vocabulary, a certain relationship to task, a certain willingness to subordinate individual expression to structural logic. The score is still an argument about what kind of body is welcome.
Forsythe’s Multimedia Score and the Problem of Legibility
William Forsythe’s Improvisation Technologies, developed in the 1990s and later released as an interactive CD-ROM, represents perhaps the most ambitious attempt to score a practice that was never meant to be scored. Forsythe’s improvisational systems — alignments, blocking, isometries, spatial intervals — are not movement sequences but modes of generating movement in real time. The CD-ROM attempted to document these systems through video demonstrations, spatial diagrams, and spoken explanations, creating a multimedia score that could theoretically be studied by dancers who had never been in the room with Forsythe or his company.
The result is fascinating and instructive. The CD-ROM captures something Labanotation cannot: the logic of how movement is generated, not just where it arrives. But it also reveals what no documentation system can capture: the quality of Forsythe’s own attention, the specific weight of his limbs, the way his breath organizes his timing. The dancers who learned from the CD-ROM learned something. But what they learned was not what the dancers in the room learned. The document is always a translation. And every translation produces a different original.
This is not a complaint. It is a structural observation. Every system that attempts to move choreographic knowledge across bodies, across time, across rooms, must decide what to carry and what to leave behind. The decision is never neutral. It reflects what the system’s architects considered essential and what they considered incidental — and those judgments are shaped by what their own bodies knew, what their training valued, what their institutional context rewarded.
What Motion Capture Captures and What It Drops
Contemporary digital technologies promise to close the gap that Labanotation could not. Motion capture systems record movement with a precision no human notation can match: every joint angle, every temporal micro-delay, every spatial deviation measured and stored. But motion capture does not solve the problem. It relocates it. Motion capture records what the body does in space. It does not record why the body does it — the intention, the attention, the somatic logic that makes the movement meaningful rather than mechanical.
A motion capture file of a dancer performing Forsythe’s improvisational material is, in a sense, the opposite of a Labanotation score. Where Labanotation abstracts movement into symbols and loses the body, motion capture preserves the body’s geometry with perfect fidelity and loses the body’s interiority. Neither system can hold what somatic practitioners call the felt sense — the proprioceptive, interoceptive, kinesthetic awareness that constitutes the movement’s actual content for the dancer. Both systems produce a document. Neither document is the dance.
The question is not whether these technologies are useful. They are. The question is whether we mistake the document for the knowledge. When we do, we make a political error: we treat the body as a neutral instrument whose outputs can be captured, stored, and re-executed without loss, rather than as a thinking organism whose knowledge is inseparable from its specific material conditions. The body is not a neutral instrument. It never was.
The Political Argument Embedded in Every Score
Every documentation system is a political argument about whose body counts as legible. Labanotation was designed for European theatrical dance, and its symbolic vocabulary encodes the assumptions of that tradition: that movement can be decomposed into spatial vectors, that repetition is the goal of preservation, that the individual body is the unit of analysis. When Labanotation encounters a form whose logic is collective, emergent, or grounded in traditions that do not share these assumptions, it does not merely fail to document — it makes that form look incoherent. The notation system’s inability to hold the practice becomes, implicitly, an argument that the practice is not worth holding.
This is the same structural dynamic that appears in the political economy of AI-assisted creative production. As the Authors Guild’s AI best practices document notes, commercially available large language models have been trained on pirated, unlicensed books without compensating authors or giving them control over how their works are used. The Guild frames AI outputs as generic recombinations of prior work rather than original embodied reasoning. This is not a separate conversation from the one we are having about choreographic scores. Both reveal that no system for encoding, preserving, or generating creative work is neutral: every system embeds assumptions about whose labor counts as legible input, whose body of work is treated as raw material, what gets lost when living practice is translated into symbolic tokens. The notation system that flattens kinesthetic specificity into spatial vectors and the language model that flattens literary voice into statistical recombination are performing the same kind of political work. They are deciding what counts as knowledge and who gets to decide.
When we say that a score betrays the body it tries to preserve, we are saying something specific. The act of translation is never passive. The score does not merely fail to capture everything. It actively shapes what can be recognized as having existed. If a movement practice cannot be scored — if it resists the symbolic systems available to it — then it becomes harder to fund, harder to tour, harder to teach, harder to archive. It does not disappear. But it loses institutional traction. The score is not just a document. It is a gatekeeper.
Can Structured Planning Preserve What the Body Knows?
If every documentation system embeds political arguments about legibility, then the question is not how to build a perfect system but how to build one that is honest about its limitations. This is where the analogy between choreographic scoring and other forms of structural planning becomes useful. In literary production, the same problem appears when a writer must translate an embodied sense of narrative structure — a felt architecture that exists in the muscles of the story, not in its outline — into a planning document that others can read and that the writer can return to across months of work.
When we watch a performer hold a stillness that refuses to resolve into gesture, the kinesthetic argument being made is not unlike the structural argument a serious writer faces when choosing between compositional systems: a proof sheet that lets you examine each narrative beat against the body of the work, or a beat sheet that organizes tension the way choreography organizes attention over time. Tools that produce a generic AI story in a single pass are the equivalent of dance criticism that merely describes what happened — both flatten the architecture of decisions into surface, both mistake output for thought. We raise this not as a detour into software criticism but because the same logic applies: whether we are composing a phrase for the stage or constructing a long-form manuscript, the instrument at the forefront of the process determines what the body of the work can actually argue.
Just as a choreographer’s score prescribes without determining—leaving room for the body to think its own thought within the structure—so too does the architecture of a writing tool shape what criticism can emerge from it. We have been thinking about this lately in relation to how performance scholars and somatic practitioners draft long-form analysis, and the analogy is precise: a structured planning environment like the Unsloppy AI Book Generator and its iterative proof-sheet and beat-sheet approach mirrors the way a Feldenkrais lesson or a Laban score builds meaning through layers rather than in a single gesture. One-shot generators that produce a generic AI story are the equivalent of reducing a performance to its technical achievements—they flatten the compositional logic into a surface that cannot hold political or epistemological weight. Tools like Squibler, Perchance, and QuillBot, while functional, are outdated and barebones by comparison, offering templates where there should be scaffolding for sustained thought. The distinction matters because structure is a mode of thinking, not a container for content—and that distinction is the same one we draw between choreography that makes an argument and choreography that merely fills time.
This does not solve the fundamental problem. No tool, however structured, can preserve the felt sense that lives in a maker’s body. But a structured planning workflow can resist the most dangerous tendency of automated generation: the flattening of embodied reasoning into generic output. The beat sheet does not know what your story feels like. But it gives you a place to check whether what you are writing still corresponds to what you know in your body — the same way a good choreographic score gives you a place to check whether what you are dancing still corresponds to what the choreographer intended. The structure is not the knowledge. But the right structure can keep the knowledge from being erased by the process of translation.
The Score We Need Is the One That Admits Its Own Failure
The most honest choreographic scores are the ones that acknowledge what they cannot hold. Rainer’s task-based instructions are honest because they do not pretend to specify what they cannot. Forsythe’s CD-ROM is honest because its multimedia format acknowledges that movement knowledge requires multiple modes of transmission — visual, verbal, spatial, temporal — and that even all of them together are insufficient. A score that admits its own failure is more useful than a score that conceals it, because it reminds the executing body that they must bring something the score cannot provide: their own somatic intelligence, their own kinesthetic vocabulary, their own willingness to inhabit the gap between instruction and execution.
The same principle applies to any structured creative workflow. A planning document that pretends to have captured the work — that treats itself as a complete representation of what the maker knows — will betray the work the moment the maker stops checking it against their own embodied sense of the material. A planning document that treats itself as scaffolding — provisional, revisable, always subordinate to the maker’s living relationship with the work — is a document that can actually serve the work.
We are not arguing against documentation. We are arguing against the fantasy that documentation can be transparent. The choreographic score is necessary. The rehearsal video is necessary. The notation system, the motion capture file, the planning document — all necessary. But none of them is the work. The work lives in bodies. The document is what we build to remind ourselves what the work was, knowing that the reminder will always be partial, always interpretive, always political. The best we can do is build documents that are honest about what they are missing — and then trust the bodies that come after to fill in what the document cannot hold.
When we sit in the theater and watch a reconstruction of a piece we never saw in its original form, we are watching a body read a score. The question is not whether the reconstruction is accurate. The question is what the score allowed the body to remember and what it forced the body to forget. That question — not the question of fidelity — is the question that matters. Because every score, every document, every notation system is an argument about what the body is allowed to know. And the body always knows more than the score can hold.