review_job_id: 8351
review_pair_id: 20749
note_path: kb/notes/irreducibility-to-rules-bounds-text-alone-not-text-plus-interpreter.md
criterion_path: kb/instructions/critique-note.md
model_partition: codex
runner: /root/closing_critique
runner_model: null
runner_effort: null
result_kind: report
outcome: null
completed_at: '2026-08-26T12:48:58+00:00'

Critique: Naur's retained passages do not establish a human-only theory bearer

Note: kb/notes/irreducibility-to-rules-bounds-text-alone-not-text-plus-interpreter.md Central commitment: Naur's retained passages establish only that coherent modification requires judgment beyond a limited criterion set, not that the necessary theory-bearer must be human, so they leave a competent text-plus-interpreter composite open. Critique mode: claim Attack outcome: partially lands

Strongest case against it

A Rylean tacit-knowledge reading of Naur rejects the note's separation of capacity from bearer as an anachronistic reconstruction. In the retained passages, Naur does not merely report that programmers happen to supply an otherwise substrate-neutral function. He says the relevance decision can be made only by “someone” who understands the whole world, says that this understanding must be contributed by the programmer, and identifies the world-knowing agent as the programmer. On the ordinary reading of the essay, possession of this situated, practical theory is what makes the person a theory-bearing programmer. Treating those clauses as a contingent implementation detail imposes a modal-proof standard on a conceptual argument whose human bearer is asserted or presupposed rather than derived.

The same opponent would reject the note's narrow gloss on “any limited set of criteria or rules.” A finite LLM-plus-artifact system is itself an end-to-end computational procedure with a limited retained state and fixed inference machinery. Calling one component a “separately competent bearer” does not show that the composite falls outside Naur's target; it assumes that the component already has the world-sensitive, program-specific know-how at issue. The note therefore has not earned its claim that Naur targets only an explicit list offered in place of judgment. At most it shows that, if some nonhuman interpreter independently possesses the relevant practical theory, attaching text to it does not remove that possession. That conditional does not establish that Naur's retained conception leaves such a bearer open on its own terms.

This attack does not fully overturn the note's narrower logical point. The retained wording does not demonstrate the modal proposition that no nonhuman system could ever possess the capacity, and the source ingest itself judges Naur's strongest in-principle claims to outrun the reported evidence. But the attack does undermine the note's stronger exegetical statements that the human-exclusivity premise is simply unstated and that the retained argument's target is merely enumerated criteria rather than a finite interpreter-mediated procedure.

How the note engages it

Partially engaged. The opening and “A text-plus-interpreter possibility has further conditions” explicitly deny that text performs judgment, deny that general language competence supplies a program theory, and require world access, program-specific acquisition, and dispositional reliability. “The argument's target bounds the inference” also recognizes that an impossibility result over the whole effective procedure would reach the interpreter, while the Scope section allows that the full essay may contain human-specific premises. These qualifications answer the easy objection that the note treats an off-the-shelf LLM as an established theory-holder.

They do not answer the strongest objection. The note offers no source-based reason for reading “limited set of criteria or rules” as only an enumerated list rather than a bounded procedure, and it moves from “the programmer contributes it” to “human exclusivity is unstated” without addressing the ordinary person-bound force of “someone” and “that is to the programmer.” Its own conditions make the proposed interpreter functionally equivalent to Naur's theory-bearing programmer, leaving the substrate question logically open but not showing that this openness belongs to Naur's retained account.

Constructive findings

  • Separate exegesis from logical reconstruction: say that the passages assert or presuppose a programmer as the actual theory-holder but do not demonstrate the stronger modal claim that only humans can instantiate the capacity.
  • Either support the restricted reading of “limited set of criteria or rules” from retained context or weaken it to a conditional: adding an interpreter escapes the criticism only if that interpreter independently supplies the world-sensitive judgment rather than merely executing the retained criteria.
  • Recast “leaves open” as external underdetermination, not as a possibility Naur affirmatively recognizes. This preserves the useful burden-of-proof point without treating a person-bound conceptual vocabulary as already substrate-neutral.

Secondary objections (optional)

  • The downstream linked notes use this claim to say that Naur's objection is “cleared” for an LLM-plus-artifact split, which is stronger than this note's own explicit caution that no current system is shown to meet the required competence and acquisition conditions.

Result: REPORT