Cunmu kills poets, not doctors

Status: Anticipated · untested

Status is derived only from the shepherd-authored triage/prediction data above -- community submissions and claims are a separate overlay and can never change it (see the participation panel below).

This is a proposed connection between two domains, generated by a language model. It is not an article and not evidence: it sits below the evidence/publication boundary. A quantitative prediction and a named kill-dataset are attached (when registered) so the claim stays falsifiable rather than merely evocative.

Claim (verbatim)

Cunmu kills poets, not doctors. The Siku quanshu zongmu's two-tier structure — titles copied into the imperial library versus cunmu, listed but not transcribed — is a natural experiment on which genres lived by state canonization and which by private demand. Being left out of the imperial transcription mattered only where the empire was the marginal reader: minor authors' literary collections had no constituency but prestige culture, so cunmu relegation was a slow death sentence; medical, agricultural, and divinatory books had buyers who never cared what the Four Treasuries thought.

Prediction clause (verbatim)

Sample cunmu titles from the zongmu in two strata — bieji (individual literary collections) and practical arts (medicine, agronomy, shushu) — and trace each in modern union catalogues of extant old editions. Primary clause (verdict follows it): the extant-today share of cunmu practical-arts titles exceeds the extant share of cunmu bieji by at least 20 percentage points. Secondary: among fully transcribed (zhulu) titles the between-genre survival gap is under 10 points — the catalogue tier interacts with genre, which a uniform-decay account cannot produce.

Kill-dataset (verbatim)

Kill: the Siku quanshu zongmu (digitized, including on Kanripo) crossed with the published Zhongguo guji shanben shumu and the online Zhongguo guji zongmu union records.

Nobody has run this test. The kill-data is named above. If you can run it — or you know the paper that already settles it — claim the kill or submit the prior. Kills and priors are credited here, by name, as they come in.

In the atlas

This conjecture is bridged, as an L1 lead, onto these Inferpedia subject pages.

Provenance

Run: Fresh agent generation · model: claude-fable-5

Generated by a fresh Fable-tier instance (claude-fable-5) at maximum effort with generation-first blindness (no repo reads, no web searches, no DB queries); 248-title list supplied at launch, titles only, no verdicts/dossiers seen; prompt pre-committed in docs/generated/conjectures_1001_wave_ledger.md and docs/generated/conjecture_fresh_fable_w04_prompt_20260708.txt. Wave W04 of docs/GOAL_CONJECTURES_1001_CAMPAIGN_20260708.md. Novelty unverified by construction. An output-token limit interrupted the first response turn before any tool call was made; the packet was still produced in a single Write with no information ingress.

Novelty / leakage triage

anticipated in the literature — this exact test has never been run

The zongmu's two-tier copied/cunmu structure is well documented (3,461 copied vs 6,793 cunmu titles), and book historians discuss commercial demand sustaining practical genres outside state canonization, but the stratified extant-today survival comparison of cunmu bieji versus practical-arts titles is un-run.

Predictions

No prediction registered yet.

Weigh in

No community feedback yet.

Add your take

Posted immediately (spam is removed). Community feedback is never an adjudicated verdict and never changes this conjecture's triage label or status above.