Answers are grounded in this paper and cite their sources.
A one-page operational protocol answering the objection that a claim of 'reading structure rather than rules' is untestable. Criterion: copy deviation - a system whose exact reproduction is indistinguishable from the original carries no self-reference (a rule-object); a system whose reproduction necessarily diverges is tracking structure. Protocol: substrate swap (numeric precision, tokenizer, prompt language, seed, context length, modality) with two measured quantities (surface divergence dS and invariant failure rate dI), classified by their ratio. Two falsifiable predictions: (P1) for a structure-tracker, dS is a monotone, directionally predictable function of substrate distance rather than seed noise; (P2) the invariant-preservation advantage over a matched rule-follower widens as substrate distance grows. Companion to The Incompleteness of Copying (DOI 10.5281/zenodo.22969175).
No takes yet. Share an insight, caveat, or question.
Xiaohei Lin (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: