Notation Systems

In the AI Story

Goodman's analysis of notation revealed that few actual symbolic systems fully satisfy the requirements. Musical notation comes close but has edge cases (grace notes, unmeasured tremolos) where syntactic differentiation fails. Literary texts satisfy the requirements at the level of spelling (each letter is a discrete, identifiable character) but break down at the semantic level, where the continuous gradations of meaning resist the sharp boundaries differentiation requires. The breakdown does not disqualify literature from the allographic category—it remains true that any text reproducing the correct word sequence is an instance of the same work—but it reveals that the allographic character of literature is less secure than the allographic character of music, because literary meaning lacks the sharp specification that musical pitch provides.

The notation/performance distinction that defines allographic arts assumes a clear division of labor: the composer specifies identity-determining features in the score; the performer fills the space the score leaves open (timbre, micro-timing, interpretive phrasing). The division holds because the notation maintains the boundary—the score says precisely which features are specified and which are left to performance. In AI collaboration, there is no comparable notation maintaining the boundary between what the human specifies and what the machine contributes. Natural-language prompts lack the formal properties Goodman required of notations—they are not syntactically disjoint (words belong to multiple categories), not differentiated (meanings have no sharp boundaries), not unambiguous (nearly every sentence admits multiple interpretations). The 'score' is therefore not a score. And the 'performance' can alter the score without detection, because no notational precision keeps them separate.

What makes this structurally consequential is that notation, in Goodman's framework, is the mechanism by which intention survives rendering. The composer's intention—this pitch, this rhythm—survives in the score with sufficient precision that performers a century later can instantiate the same work. The precision depends on the notation's formal properties. When those properties are absent—when natural language replaces formal notation as the medium of specification—the intention's survival across the rendering process is not guaranteed. The machine may render what the human intended, or it may render a statistically probable neighbor of what the human intended, and the difference may be invisible to both parties. The notational precision that would make the difference detectable is structurally unavailable in natural-language collaboration.

Origin

Goodman's theory of notation occupies Languages of Art, Chapter IV, titled 'The Theory of Notation.' The framework was developed in response to the question of what makes music and literature allographic—what formal properties enable a work's identity to be preserved across performances and copies. Goodman's answer was that a notation system must satisfy the five requirements, and that the requirements are jointly necessary and sufficient for allographic status. The analysis has become foundational in philosophy of music, where it shaped debates about authentic performance, the ontology of musical works, and the relationship between scores and improvisational practices. The framework has proven unexpectedly applicable to digital technologies, where the distinction between notated and non-notated arts maps onto the distinction between symbolically specified (code, formal languages) and symbolically dense (images, natural language) artifacts.

Key Ideas

Five strict requirements. Notation demands syntactic and semantic disjointness, differentiation, and unambiguity—jointly necessary for preserving work-identity across instances without relying on productive history.

Notation makes allography possible. Only symbol systems satisfying the notational requirements can support allographic arts, where correct instantiation suffices for work-identity regardless of who performs or copies.

Natural language is not notation. Prompts lack syntactic differentiation, semantic precision, and unambiguity—the formal properties that would maintain boundaries between human specification and machine contribution.

Intention's survival requires notation. Without notational precision, the human's worldmaking intention may not survive the machine's rendering intact—the gap where specification becomes interpretation is structurally undetectable.