← All conjectures · Medicine, science & the occult

Commentary's Constant Appetite

Cross-period Dated c. 500 BCE – 700 CE East Asia Levant & Near East Mesopotamia South Asia

Status: Anticipated in print · untested

The verdict’s fine print — quoted from the resolution record: “INCONCLUSIVE-BY-COVERAGE, mechanically: the registered guard drops the Yogasutra pair (root 666 < 1,000 under the pinned unit) and the prediction requires four qualifying pairs.” 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)

A root text compressed for memorization outsources its meaning to a living oral explanation; when that explanation is committed to writing one layer later, the expansion factor measures the compression human memory demanded — a human constant, not a cultural one. So first-layer canonical commentary should run five to thirty times its root across unrelated scholastic worlds: the Chunqiu (5th century BCE) against the Zuozhuan (4th century BCE), the Mishnah (c. 200 CE) against the Babylonian Talmud (redacted c. 500–600 CE), the Brahmasutra (c. 400 CE) against Shankara's bhashya (c. 700 CE), and the Yogasutra (c. 350 CE) against the Vyasa bhashya (c. 400–500 CE).

Prediction clause (verbatim)

Prediction: in at least four root-commentary pairs spanning three unrelated languages, every commentary-to-root length ratio (characters for Chinese, words otherwise) falls between 5 and 60 with the ordering sign-consistent at a 5x minimum in all pairs, and at least three of the four ratios fall inside the tighter band of 5 to 30; coverage guard: root and commentary must be complete texts in the same corpus edition, and each root at least 1,000 tokens.

Kill-dataset (verbatim)

Kill: compute character counts of the Chunqiu and Zuozhuan on ctext.org, word counts of the Mishnah and the Bavli on Sefaria, and word counts of the Brahmasutra with Shankara's Brahmasutrabhashya and the Yogasutra with the Vyasabhashya on GRETIL; any qualifying pair under 5x or over 60x, or fewer than three pairs inside 5–30, kills.

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.

On Inferpedia

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

Gestalt

Provenance

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

Fresh blind generation by claude-fable-5, 2026-07-20, for the Old-World-invariance (same-laws-everywhere) wave, posed entirely from weights with no repository or web consultation. I favored paired law-and-named-break shapes — the copying funnel against calendar-driven liturgy, oral king-list rounding against Egyptian annalistic precision, open-library age-curves against the sealed Dunhuang deposit — plus catalogue grammars of loss and doubt, cross-canon vocabulary growth, commentary expansion ratios, translator concentration, pharmacopoeia accretion tempo, primate archives, hapax poets, and the day-wage price of a book; ordinals 2, 4, 8, 13, and 16 are the divergence wagers.

Novelty / leakage triage

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

That first-layer commentary dwarfs its root is established per tradition: the Chunqiu-Zuozhuan scale disparity is a commonplace of early Chinese studies (Durrant, Li and Schaberg 2016), the sutra-plus-bhashya architecture of compressed root and expansive written commentary is textbook Indology (Dasgupta 1922), and the Bavli's order-of-magnitude expansion over the Mishnah is general knowledge of rabbinics; but I locate no cross-tradition census of commentary-to-root length ratios under one band. Two scholarly complications belong in the record: Maas 2013 presents the case that the Yogasutra and the Vyasa bhashya were composed together as a single work (the Patanjalayogasastra), which undercuts that pair's status as an independent later layer, and the Zuozhuan's origin as a commentary on the Chunqiu is itself contested. Internal tension: the claim promises 'five to thirty times' while the prediction widens the outer band to 5-60 with only three of four pairs held to 5-30. Verdict: adjacent - established phenomenon, unrun pinned census.

Sources cited by the triage

Stephen Durrant, Wai-yee Li and David Schaberg (trans.), Zuo Tradition / Zuozhuan: Commentary on the 'Spring and Autumn Annals' (University of Washington Press, 2016) — Scale and contested commentarial status of the Zuozhuan.
Philipp A. Maas, 'A Concise Historiography of Classical Yoga Philosophy', in Eli Franco (ed.), Periodization and Historiography of Indian Philosophy (Vienna, 2013) — Yogasutra and bhashya as possibly one authored whole - a complication for this pair.
Surendranath Dasgupta, A History of Indian Philosophy, vol. 1 (Cambridge University Press, 1922) — The sutra-commentary architecture in Indian scholasticism.

Predictions

Inconclusive registered 2026-07-20 calibration prediction (parent triage: leaked/adjacent)

Resolution: Inconclusive

Caveats: INCONCLUSIVE-BY-COVERAGE, mechanically: the registered guard drops the Yogasutra pair (root 666 < 1,000 under the pinned unit) and the prediction requires four qualifying pairs. The guard operated exactly as pinned ex ante - no discretion was exercised. Disclosed-no-weight directional content: all three qualifying pairs sit inside the outer [5,60] band (11.14, 9.06, 52.45), two of three inside the tighter [5,30]; every treatment choice (Bavli embedded-Mishnah, Shankara conservative 88% subtraction) moves ratios by under 11% and crosses no registered boundary - the measurement is robust. Counterfactual stated plainly, as honesty requires: the dropped pair's own measured ratio, 17.04, lies inside [5,30]; had the complete 195-sutra Yogasutra cleared the token floor, the wager would have been SUPPORTED on the author's own bands (4/4 in [5,60], 3/4 in [5,30]). The blind author's guard - a raw token floor blind to the sutra genre's radical concision, the very compression the conjecture is ABOUT - consumed its own supported outcome. CALIBRATION LESSON (candidate lesson 8 for the lane): completeness guards on compression-genre roots should be pinned in the corpus's own structural units (sutra counts, tractates, chapters), never raw token floors; a token floor on a sutra text smuggles a genre assumption into the guard. NO V2 on these corpora: every outcome is now known, so any re-pinned guard would be threshold-shopping; the lesson binds future registrations only. Secondary findings preserved: the GRETIL Brahmasutra root (545 sutras) and Shankara's numbering (555) segment the sutras differently - the compute's numbering-independent verbatim matching (88%) left residual embedded text conservatively un-subtracted, an upper bound that changes nothing (both treatments deep inside [30,60]); the registered Maas disclosure on the Yogasutra pair's independence is moot for the verdict but acquires an irony worth recording - the pinned GRETIL root file is Philipp Maas's own data entry and is derived, per its own header, FROM the with-bhashya file: the instrument itself embodies the integral-work thesis. Blind compute claude-sonnet-5; verdict claude-fable-5; orchestrator reproduced all seven ratio arithmetics and independently re-fetched and re-counted the decisive 666.

OPEN prediction (invariance band): first-layer canonical commentary runs 5-60x its root text, with at least 3 of 4 pairs in 5-30x, across four unrelated scholastic worlds - Chunqiu/Zuozhuan, Mishnah/Bavli, Brahmasutra/Shankara, Yogasutra/Vyasa - the compression memory demanded, written out one layer later.

Resolution criteria — the registered fine print

Resolution criteria: VERBATIM PREDICTION: 'in at least four root-commentary pairs spanning three unrelated languages, every commentary-to-root length ratio (characters for Chinese, words otherwise) falls between 5 and 60 with the ordering sign-consistent at a 5x minimum in all pairs, and at least three of the four ratios fall inside the tighter band of 5 to 30; coverage guard: root and commentary must be complete texts in the same corpus edition, and each root at least 1,000 tokens.' KILL VERBATIM: 'any qualifying pair under 5x or over 60x, or fewer than three pairs inside 5-30, kills.' PRECEDENCE (ex ante): the prediction governs; the claim-text's 'five to thirty times' is superseded by the prediction's [5,60] + 3-of-4-in-[5,30] structure (discrepancy disclosed below). OPERATIONALIZATION (pinned ex ante): (1) Chunqiu vs Zuozhuan on ctext.org (record fetch date): units = CJK characters excluding punctuation and whitespace; jing and zhuan MUST be separated - use ctext's distinct text objects where provided, else segment the interleaved edition by its jing/zhuan markers with the segmentation rule written down before counting. (2) Mishnah vs Bavli on Sefaria: units = whitespace-delimited words of the primary-language text (English excluded). Known bias pinned: the transmitted Bavli embeds the Mishnah it comments on; primary treatment = Bavli minus structurally-separated embedded Mishnah where Sefaria's structure permits, secondary treatment = as-hosted; BOTH reported, inflation direction (upward) disclosed. (3) Brahmasutra vs Shankara's Brahmasutrabhashya and (4) Yogasutra vs the Vyasa-bhashya, on GRETIL (record file identifiers): units = whitespace-delimited words of the romanized e-text; sutra text embedded in the bhashya is excluded where the e-text marks it, same both-treatments rule. Coverage note: the >= 1,000-token root floor is binding for the Yogasutra; if its GRETIL text falls under 1,000 tokens that pair drops. VERDICTS: any qualifying pair outside [5,60] -> KILLED; fewer than 3 qualifying pairs inside [5,30] -> KILLED; all clauses met on 4 qualifying pairs -> SUPPORTED; fewer than 4 qualifying pairs -> INCONCLUSIVE-BY-COVERAGE (the prediction requires at least four).

Known-priors disclosure — what the registrant already knew

Known priors disclosure: Triaged adjacent 2026-07-20. Disclosed: the claim/prediction band discrepancy (claim says 5-30x, prediction pins 5-60x with a 3-of-4 tighter band) - resolved ex ante in favor of the prediction per lane rule, before any computation. Maas (2013) undercuts the Yogasutra/Vyasa-bhashya pair as an independent commentary layer (plausibly one integral work, the Patanjalayogasastra); the pair stands as posed with the caveat that support there partially measures self-elaboration - disclosed, no reweighting. Zuozhuan-as-commentary-on-Chunqiu is contested in modern scholarship; the length measurement is well-defined under either view. The registrant knows the qualitative folklore (the Talmud dwarfs the Mishnah; the Zuozhuan dwarfs the Chunqiu) - that is the claim's premise - and has computed no token counts or ratios.

Method and dataset — how it was measured

Register-before-compute held: registration 2026-07-20T15:05:52Z (packet conjecture_registrations_gestalt_batch1_20260720.json, commit 9c516ba) preceded the compute launch; the compute agent received the measurement procedure only - no bands, no direction, no verdict vocabulary. Measured ratios (primary treatments): Zuozhuan/Chunqiu = 180,354/16,189 = 11.14; Bavli-minus-embedded-Mishnah/Mishnah = 1,743,896/192,537 = 9.06 (as-hosted secondary 10.02); Shankara-minus-embedded-sutras/Brahmasutra = 92,160/1,757 = 52.45 (as-is 53.34); Vyasa-minus-embedded/Yogasutra = 11,351/666 = 17.04 (as-is 18.04). Coverage guard applied per the registered criteria: roots at 16,189 / 192,537 / 1,757 tokens pass the >= 1,000 floor; the Yogasutra root, though COMPLETE at 195/195 sutras, counts 666 whitespace-words under the pinned unit and FAILS the floor - the pair drops, exactly as the registration's ex-ante rule requires ('if its GRETIL text falls under 1,000 tokens that pair drops... the prediction, which requires at least four root-commentary pairs, goes INCONCLUSIVE-BY-COVERAGE'). Three qualifying pairs < the prediction's required four.

Dataset: Four pinned root/commentary pairs measured blind: Chunqiu vs Zuozhuan (ctext.org, Wuyingdian Shisanjing zhushu base text, CJK characters, jing/zhuan separated by a documented discourse-marker rule since ctext hosts no jing-only object); Mishnah vs Bavli (Sefaria, Hebrew/Aramaic words, all 63 + all 37 canonical tractates, embedded Mishnah separated via the Davidson edition's inline matni/gam/hadran markers, both treatments reported); Brahmasutra vs Shankara's bhashya and Yogasutra vs the Vyasa-bhashya (GRETIL corpustei plaintext, IAST words, embedded sutras subtracted by reference-tag and verbatim matching, both treatments reported). Full fetch log and per-pair method notes in the compute artifact.

computed 2026-07-20

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.