Nanopublication — Detecting AI prose maps cleanly onto the given/instituted split — until the 'borrowed codex' has no borrower

Context

ⓘ Thin page — the source itself is thin (teaser, notice, paywalled preamble). Distinct from truncation: nothing was cut by the pipeline.

What the source says ^[1]

A Hyperallergic newsletter mixes several short news items with one substantive piece: editor Valentina Di Liscia argues that art-world writing (wall labels, press releases, artist statements) increasingly shows the fingerprints of AI generation — a recognizable cadence ('It's not just X, it's Y'), overconfident in tone but hollow in content — and demonstrates this by having Claude write a sample paragraph that she then dissects. The rest of the digest (MIT professors decorating surveillance cameras, an Art Basel Paris lockdown, a V&A strike) is unrelated news.

Where the lens grips

P3a's given/instituted distinction grips the editorial detection act precisely: what the editors read is only the given surface — rhythm, phrasing, that recognizable cadence — and from that surface alone they infer that no instituted rule, no singular 'why these forms', sits behind it. P21's claim that 'fashion is the name we give to borrowed codices at scale' also grips: the piece frames AI prose's detectability as its formulaic, recyclable sameness, the opposite of a built, singular codex.

Where it breaks

P21 slips IF one reads 'borrowed codex' strictly — a second party relaying an identifiable first party's wave, as in plagiarism or pastiche of a named style. The Claude paragraph borrows from no single codex: it is a statistical average across thousands of texts, instituted by no one in particular. It holds only if 'codex' is loosened to mean any detectable regularity whatever its source, which P21's own language of borrowing and relaying does not license — the proposition presupposes a prior, singular style to take from. P19a is similarly strained: it is built for an invariant the artist cannot yet explain, not for a text with no artist behind it at all, so the 'mute intentional layer' it describes and the 'no layer to begin with' of the AI case are not the same condition, though the surface symptom (ricochet cannot complete) looks identical.

Propositional grid ^[2]

Read this grid. “Not evaluated” does not mean the source has nothing to say about that proposition. It means the reading did not instruct it. That is not a result, and it counts in no denominator. A silence, by contrast, is a finding: the reading touched the proposition without concluding, or noted its absence itself.

Evaluated

What was instructed. These propositions have been judged and carry a current statement: it is that statement the reading tested, not the founding one.

• P3a — judged in Note 05 · current statement : say GIVEN invariant and INSTITUTED invariant, never 'physical' and 'intentional' — the axis is MODE OF EXISTENCE, read as regime of access: the given is what a sensor measures on the surface WITHOUT the codex; the instituted is the rule that elects relations in that surface and that only the disclosed codex deciphers

Not evaluated (34) — P1, P2, P3, P4, P5, P6, P7, P8, P9, P10, P11, P12, P13, P14, P14a, P14b, P15, P16, P16a, P17, P18, P18a, P19, P20, P22, P23, P24, P25, P26, P27, P28, P29, P30, P31

Overall verdict: Bounded — taxonomy of the August 2026 pass, predating the move to relational verdicts. Kept as metadata: the gap between the two forms is data, not residue. This reading’s judgement is in the grid above.

Reading of 2026-10-22 · model claude-sonnet-5 · source 6,275 chars · 1 attempt(s)