← All conjectures · South Asian text cultures

The examples stopped changing

Medieval Dated c. 1000 CE – 1400 CE South Asia

Status: Anticipated in print · untested

The verdict’s fine print — quoted from the resolution record: “INCONCLUSIVE BY REGISTRATION DEFECT, transparently recorded: the conjecture's own primary clause takes as denominator the SD illustrative verses THAT RECUR ANYWHERE (excluding Vishvanatha's self-composed, never-recurring examples); the registration mis-operati…” Read the full caveats ↓

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 conjecture imagined by a language model — drawn from its trained weights and held to falsifiability, novelty, and plausibility, not to any one method: it may join two or more fields, or none. 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.
“Anticipated in print” means the direction of this claim already appears in scholarship, while this exact test has never been run — it records prior art, not proof; nothing here counts as supported or falsified until a registered resolution says so.

Claim (verbatim)

Every Sanskrit poetics manual illustrates its definitions with example-stanzas, and a manual could in principle draw fresh examples from the living poetry of its own day. After Mammata's Kavyaprakasa (eleventh century) they largely stopped doing so. The Kavyaprakasa's stock of illustrations became a canon in itself, recopied by the later handbooks, Visvanatha's Sahityadarpana, the Candraloka and their kin, rather than replaced by new sampling. The consequence for the record is severe. Once the illustration-canon froze, the manuals ceased to be a window onto the poetry actually circulating in the twelfth through seventeenth centuries, and any poet who arrived after the freeze is invisible to the tradition's own examples. The recycling ratio measures the freeze directly, and its height means that the huge apparatus of quoted verses in late poetics is not a broad census of kavya but a single eleventh-century snapshot passed hand to hand, with famous exceptions such as Jagannatha's Rasagangadhara illustrating from his own compositions that prove the rule by breaking it. What looks like a wide sampling of poetry is a narrow inheritance wearing many covers.

Prediction clause (verbatim)

Prediction: matching the illustrative stanzas of the Sahityadarpana against those of the Kavyaprakasa by pada-level string comparison on the GRETIL and SARIT e-texts, more than 40 percent of the Sahityadarpana's illustrative verses that recur anywhere will be inherited from the Kavyaprakasa's example-stock rather than drawn from post-Mammata poetry; the greater-than-40-percent inheritance share is the deciding clause. Only verses functioning as illustrations and not doctrinal karikas are counted, an author's self-composed examples are tallied separately as the control, and the test voids for coverage if fewer than 150 illustrative verses can be isolated in either manual.

Kill-dataset (verbatim)

Kill: the GRETIL and SARIT e-texts of Mammata's Kavyaprakasa and Visvanatha's Sahityadarpana, comparing their illustrative-verse sets pada by pada, with the verse-indexes of the standard editions (for instance Jhalakikar's Kavyaprakasa) as the reconciliation control.

Browse the registry: Digitized texts →

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 scholarship. Kills and prior scholarship are credited here, by name, as they come in.

How this item is catalogued

Catalogue tags are curatorial metadata that make the conjecture findable. They are not findings, not a novelty verdict, and naming a source family is not a promise that the data exist or are accessible. Each dimension below says whether it was assigned, reviewed with none applying, uncertain, or not yet reviewed, and who recorded it.

Place & era tags — Assigned: South Asia; Medieval Tagged by claude-fable-5
Dating — Phenomenon: Dated c. 1000 CE – 1400 CE Mammata's Kavyaprakasa (11th c. CE) and its illustrative-verse canon, recycled by Visvanatha's Sahityadarpana (14th c.) rather than resampled from contemporary poetry. Tagged by claude-sonnet-5 + claude-fable-5 (gate)
Subject — Assigned: Literature & poetics; Texts, scribes & transmission a poetics-manual illustration-canon fossilizing after Mammata, ceasing to sample contemporary poetry Tagged by claude-sonnet-5
Finer subject topic — Reviewed — none applies: No listed page is squarely about the example-stock of poetics manuals. Tagged by claude-opus-5 draft (requested max effort) adjudicated by claude-fable-5-1 driver, lane 1da5; runtime not independently attested
Claim level — Assigned: Record claim The prediction is the share of one manual's illustrations inherited from its predecessor, i.e. what later handbooks actually copied. Tagged by claude-opus-5 draft (requested max effort) adjudicated by claude-fable-5-1 driver, lane 1da5; runtime not independently attested
Method — Reviewed — none applies: An inheritance share above 40 percent of recurring illustrative verses; a share comparison, with the self-composed examples tallied only as a side control. Tagged by claude-opus-5 draft (requested max effort) adjudicated by claude-fable-5-1 driver, lane 1da5; runtime not independently attested
Evidence source — Assigned: Digitized texts; Published editions Kill test compares the illustrative-verse sets of the Kavyaprakasa and the Sahityadarpana pada by pada on the GRETIL and SARIT e-texts, with the standard editions' verse-indexes as reconciliation control. Tagged by claude-opus-5 draft (requested max effort) adjudicated by claude-fable-5-1 driver, lane 1da5; runtime not independently attested
Research stage — Test resolved
Dataset family of the resolution — Assigned: external the resolution loaded a named external corpus or database, not the project's manuscript census Tagged by claude-fable-5-1 (driver, lane 1da5; requested max effort, runtime not independently attested)

On Inferpedia

This conjecture is linked to the following pages on Inferpedia, an encyclopedia of the missing — working atlas pages, some still early scaffolding.

Provenance

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

Fresh blind generation by claude-fable-5, 2026-07-17, Sanskrit anthology/citation wave instrument-anchored on GRETIL and SARIT (the inst-gretil-sarit registry row is live), with the printed kavi-indexes, Sternbach's Descriptive Catalogue of Poets Quoted in Sanskrit Anthologies and Inscriptions / Maha-subhasita-samgraha, Kosambi & Gokhale's Harvard SRK, Kosambi's Bhartrhari edition, and the New Catalogus Catalogorum as controls. Open lane = the anthology-and-citation geometry of classical kavya and sastra, held disjoint from the 75 prior south_asia_text_culture items across w14-southasia (40), breadth-india (17) and breadth-india-w2 (18); every candidate name grepped diacritic-insensitively against all three packets before finalizing. Instruments may be reused under a new operation (e.g. the Abhinavabharati serves item 7's positional-concentration test, distinct from C#6's Kohala divergent-fragment test). Dropped for prior-wave overlap: Rajatarangini pre-Kalhana predecessor citation-audit (B#13 already uses Suvrata / Ksemendra's Nrpavali / Helaraja's Parthivavali as its evidence); Ksemendra attested-vs-extant oeuvre (operation duplicates B#14's Abhinavagupta one-third-lost, and Ksemendra is used in B#4/B#13); Brhatkatha / Gunadhya (B#4 already posed the stemma); Rajasekhara Kavyamimamsa poet-census (shares its primary text with C#5, judged too close); lost-nataka census via Natyadarpana / Srngaraprakasa / Natakalaksanaratnakosa (B#5 owns the dramaturgy playbill); dharmasastra nibandha digest-fragments e.g. reconstructed Brhaspatismrti (A#6 owns digest-eats-smrti); vrajya / deity-section per-class survival (mechanism duplicates C#18's Paripatal deity-class survival); Carvaka / philosophy refutation-survival (C#4 plus the owned Mimamsa/Buddhist-refutation ground).

Novelty / leakage triage

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

That the Kavyaprakasa's illustrations became canonical and that Visvanatha leans on Mammata is a commonplace of the histories, and Jhalakikar's edition traces the example-verses' sources, so the nearest prior art is thick. But the pada-level inheritance share between the two example-stocks has never been computed. The 40-percent clause could genuinely fail, because Visvanatha conspicuously illustrates with his own and his family's compositions - a practice the histories note - which drains the recurring-verse pool the clause is defined over. Un-run arithmetic over published texts.

Sources cited by the triage

V. Jhalakikar (ed.), Kavyaprakasa of Mammata with the Balabodhini, 7th ed. (Poona, 1965), verse-source apparatus
P. V. Kane, History of Sanskrit Poetics, 4th ed. (Delhi, 1971)
S. K. De, History of Sanskrit Poetics, 2nd ed., 2 vols. (Calcutta, 1960)

Predictions

Inconclusive registered 2026-07-18 calibration prediction (parent triage: leaked/adjacent)

Resolution: Inconclusive

Caveats: INCONCLUSIVE BY REGISTRATION DEFECT, transparently recorded: the conjecture's own primary clause takes as denominator the SD illustrative verses THAT RECUR ANYWHERE (excluding Vishvanatha's self-composed, never-recurring examples); the registration mis-operationalized this as ALL SD illustrative verses. Under the registered (defective) reading the computed shares (0.155/0.199) fall far below 0.40 - but the conjecture's own reading is not decided: if under ~39% of SD's examples recur anywhere (plausible, given Vishvanatha's well-known habit of quoting his own poetry), the inherited share among recurrers could still exceed 0.40. Killing on the registrant's misreading would be a false kill. PATH TO RESOLUTION: re-register with the conjecture's denominator verbatim and compute recurrence against a wider corpus (GRETIL anthologies + alamkara corpus) to delimit the recurring subset; the present compute's per-verse match table (in the artifact) is reusable. SYSTEM LESSON recorded: registrations must quote the prediction's denominator verbatim, not paraphrase it.

Registered against the GRETIL e-texts of Mammata's Kavyaprakasha and Vishvanatha's Sahityadarpana. Claim under test (primary clause): MORE THAN 40% of the Sahityadarpana's illustrative verses are inherited at the pada level from the Kavyaprakasha's example corpus.

Resolution criteria — the registered fine print

Resolution criteria: DATA: GRETIL plain-text editions of both works (freeze URLs + sha256). POPULATION: illustrative verses (udaharana stanzas, as distinct from karika/vrtti prose) in each work, identified by verse-boundary markup in the e-texts; where markup is ambiguous, a verse counts as illustrative if it is quoted (not numbered as a karika of the treatise itself). MATCH: a Sahityadarpana example verse counts as inherited if at least one full pada (quarter-verse, >= 8 aksharas as transliterated) occurs verbatim (normalized for sandhi-neutral orthography: case-insensitive, punctuation/whitespace-stripped) in the Kavyaprakasha's example corpus. R = inherited / total SD example verses. CLAUSE PRECEDENCE: (1) INCONCLUSIVE_BY_DESIGN if either e-text lacks recoverable verse boundaries or the SD example population cannot be delimited to within 20%; (2) SUPPORTED if R > 0.40; (3) REFUTED if R <= 0.40. Compute firewalled from thresholds and direction.

Known-priors disclosure — what the registrant already knew

Known priors disclosure: Shepherd triage (2026-07-17) graded this adjacent: Jhalakikar's apparatus and Kane document KP-dependence qualitatively, but the pada-level share is un-run; Vishvanatha's self-composed examples (he showcases his own poetry) push against the threshold, so the registrant regards the outcome as genuinely uncertain - if anything leaning REFUTED. The compute agent is firewalled from this disclosure.

Method and dataset — how it was measured

Register-before-compute (conjecture_prediction_batch_20260718.json row 3; prior DISCLOSED: genuinely uncertain, leaning refuted). Firewalled Sonnet compute (no thresholds/direction). computed_at postdates registered_at (rule 6).

Dataset: GRETIL e-texts (sha256-frozen): Kavyaprakasha corpustei XML (609 illustrative verses extracted - including the ~86% inlined in curly-brace prose, invisible to naive markup extraction) and Sahityadarpana -comm XML (665 illustrative verses, +-8-10%, three authorial layers separated by editorial markers). Sliding 16-char normalized-window pada matching.

computed 2026-07-18

Weigh in

No community feedback yet.

New here? Create an account first

Create an account or sign in and your feedback is tied to you — you can track it, get replies, and claim this conjecture so others know you’re working on it. Prefer not to? Just leave your take below as a guest — only the name you type is shown.

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.