Rightness of Rendering

In the AI Story

Goodman's concept of rightness replaced an entire philosophical tradition that evaluated representations by their fidelity to reality. The correspondence theory of truth—a statement is true when it corresponds to the facts—presupposes that facts exist independently of the statements that describe them. Goodman's nominalism and constructivism rejected this presupposition: facts are version-relative, and what counts as a fact depends on which version one is operating within. The physicist's facts and the historian's facts are constituted by different symbol systems, answerable to different standards, and there is no neutral standpoint from which to adjudicate between them. Rightness provides the evaluative framework that correspondence cannot: a version is right when it satisfies the criteria internal to the worldmaking project of which it is a part.

The criteria of rightness are multiple and must be satisfied simultaneously. A version that achieves coherence but lacks productivity (it holds together but reveals nothing new) is less right than a version that achieves both. A version that fits with other accepted versions but fails to serve its stated purposes is defective by Goodman's standards. The evaluation of rightness is therefore not a single test but a multi-dimensional assessment, and the dimensions can conflict—a version might be highly productive while failing to fit smoothly with prior versions, forcing a choice between conserving the old framework and adopting the productive innovation. Goodman did not provide an algorithm for resolving such conflicts. He provided the vocabulary for recognizing that the conflict is genuine, that both dimensions matter, and that the resolution requires judgment rather than calculation.

In the context of AI-generated outputs, rightness of rendering becomes the central diagnostic question. An AI-generated passage may be internally coherent (each sentence follows logically from the previous), may fit with conventions of the genre (academic prose, business writing, philosophical argument), and may serve the stated purpose (explaining a concept, making a case). But if the passage fails the productivity test—if it rehearses familiar points without generating new understanding—or if it fails the standards-compliance test in subtle ways (deploying philosophical references that do not actually support the claim, as in the Deleuze fabrication Segal caught), then the passage is wrong despite its surface competence. The wrongness is often invisible to the person who prompted it, because the smooth coherence of the output satisfies the immediate criteria while concealing the failure of the deeper ones.

The preservation of rightness in AI-augmented creative work requires that the human worldmaker maintain the capacity for multi-dimensional evaluation—asking not merely 'Does this cohere?' but 'Does this fit? Does this produce understanding? Does this meet the standards my project demands?' The questions cannot be outsourced to the rendering engine, because the rendering engine optimizes for surface properties (coherence, conventional fit) without access to the worldmaking purposes that determine whether the surface properties are right. The human must hold the purposes, evaluate the rendering against them, and reject technically competent outputs that fail the rightness test. The discipline is difficult precisely because competence and rightness look identical from the outside, and distinguishing them requires the kind of judgment that only a purposeful worldmaker, grounded in lived experience and answerable to standards she has internalized through practice, can exercise.

Origin

The concept appears fully formed in Languages of Art (1968), where Goodman argued that the traditional question 'Is this representation accurate?' should be replaced with 'Is this rendering right?' The substitution was not merely terminological—it was a philosophical reorientation that displaced the metaphysics of copy-fidelity with the pragmatics of functional success. A rendering succeeds when it does the work it was designed to do, and the work is determined by the worldmaking project, not by a standard of correspondence that presupposes a world independent of all versions. The concept was refined across Goodman's later works, particularly Ways of Worldmaking and his essays collected in Of Mind and Other Matters (1984), where he extended the framework from aesthetics into epistemology generally.

Key Ideas

Rightness is multi-dimensional. Coherence, fit, productivity, and standards-compliance must be satisfied simultaneously—no single criterion is sufficient, and dimensions can conflict.

Purposes ground evaluation. Whether a rendering is right depends on the purposes of the worldmaking project—and purposes require a purposeful worldmaker, not a rendering engine.

Competence is not rightness. Technically correct, conventionally appropriate, internally coherent outputs can fail the rightness test by lacking productivity or failing subtle standards the surface properties conceal.

Judgment is irreducible. Evaluating rightness requires the multi-dimensional, purpose-grounded, standards-internalized judgment that only a human worldmaker engaged with the material can provide.