← All conjectures · Jewish book cultures

A second Talmud hiding inside the first

Iron Age & Antiquity Dated c. 200 CE – 400 CE Levant & Near East

Status: Falsified

The verdict’s fine print — quoted from the resolution record: “(a) The QUALITATIVE direction survives: R = 0.8719 < 1 (pooled cross-check 0.8856, same direction), i.e.” 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.

Claim (verbatim)

Within the Palestinian Talmud the three "gates" of the order Nezikin - Bava Kamma, Bava Metzia, Bava Batra - do not read like their neighbours. Saul Lieberman argued in Talmud shel Kisarin (1931) that this Yerushalmi Nezikin is a separate, earlier redaction produced at Caesarea rather than Tiberias: its sugyot are markedly terser, its cited sages skew to an earlier generation, and its technical-redactional vocabulary diverges from the rest of the Yerushalmi. The loss framing is this: what we call "the Yerushalmi" is not one book but a survival-composite, and one of its members is a fossil of a Palestinian talmudic tradition - the Caesarean - that otherwise vanished entirely. If Lieberman is right the seam is measurable, because two redactional workshops leave two statistical signatures in the same manuscript. Mechanism: editorial habits (sugya length, discourse markers, the roster of named tradents) are fingerprints of a redaction, and they do not change at an order-boundary unless the redaction itself changes. Prediction restated: the mean segment length of Yerushalmi Nezikin runs far below the Yerushalmi average, and its distribution of discourse-marker terminology diverges sharply - a quantitative seam at exactly the Bavot, marking the buried Caesarean Talmud that survives nowhere else.

Prediction clause (verbatim)

Prediction: measuring mean segment (sugya-unit) character length across Sefaria's machine-readable Talmud Yerushalmi, the three Nezikin tractates Bava Kamma, Bava Metzia and Bava Batra will show a mean segment length below two-thirds (a ratio < 0.67) of the mean across the other extant tractates, and at least one high-frequency redactional discourse marker will differ in per-segment rate by a factor of two or more between Yerushalmi Nezikin and the rest - primary clause: the < 0.67 Nezikin-to-rest mean-segment-length ratio; the verdict follows it. The test voids for coverage if the Bavot segmentation cannot be aligned to the rest of the corpus.

Kill-dataset (verbatim)

Kill: Sefaria's open Talmud Yerushalmi text - compute per-tractate mean segment length and marker-term frequencies, isolating Bava Kamma / Metzia / Batra against the remaining tractates, with the Academy of the Hebrew Language Yerushalmi (Leiden MS) as the critical control and Lieberman's Talmud shel Kisarin as the philological baseline.

Browse the registry: Digitized texts →

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: Levant & Near East; Iron Age & Antiquity Tagged by claude-fable-5
Dating — Phenomenon: Dated c. 200 CE – 400 CE Saul Lieberman's thesis (Talmud shel Kisarin, 1931) dates the Yerushalmi Nezikin 'Bavot' tractates to an earlier, separate Caesarean redaction, standardly placed in the 3rd-4th century CE. Tagged by claude-sonnet-5 + claude-fable-5 (gate)
Subject — Assigned: Philology, editions & stemmatics textual-critical argument that part of a composite work is a separate earlier redaction Tagged by claude-sonnet-5
Finer subject topic — Reviewed — none applies: A redactional seam inside the Yerushalmi is not squarely any listed subject page. 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 a measurable redactional seam inside a transmitted Talmud, a property of how the composite text was made. 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: Registered statistic: a mean segment-length ratio and a discourse-marker per-segment rate ratio; a stylometric seam, not a stemma or a fitted distribution. 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 computes per-tractate mean segment lengths and marker rates on Sefaria's Yerushalmi with the Academy's Leiden-MS text as the critical 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, Jewish text-culture wave instrument-anchored on the open Sefaria corpus and its cross-reference link data, with standard critical editions as controls: every kill names a real corpus and a countable operation (coverage maps, citation-formula counts, link-orphan shares, digest-fraction, recension divergence, citation-decay), thresholds far from 1 with explicit coverage guards distinguishing what Sefaria holds from what existed. Ground is CITATION-GEOMETRIC and disjoint from the 2026-07-10 w19 Jewish wave, which was material-culture-of-loss (colophons, parchment, genizah, masora, binding fragments): no material-culture re-posing here. Two candidates were dropped after a grep of all fresh packets: a Tosafot-density-by-tractate survival item (pre-empted by w02-philosophy #18, which already correlates per-folio Tosafot density with manuscript survival), and a piyyut liturgy-vs-anthology survival item (pre-empted by w01-literature #28 'Liturgy out-survives fame', keyed to Davidson's Thesaurus). The Mishneh Torah item here tests EXTERNAL source-attribution erasure vs the Tur (loss of source-geometry), deliberately distinct from minds-w02 #25, which tests the code's INTERNAL recall/promise cross-references. Two items are marked Kill (not yet built) where the deciding corpus (Kohut's Arukh apparatus; Lewin's parallel Iggeret recensions) is not yet digitized.

Novelty / leakage triage

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

Lieberman's Talmud shel Kisarin (1931) states the qualitative result the conjecture rests on - the Bavot's sugyot are markedly terser, their tradents earlier, their redactional terminology divergent - and the distinctness of Yerushalmi Neziqin remains consensus through Sussmann's re-examination and the Rosenthal-Lieberman Escorial edition, whatever the verdict on Caesarea. But no one has computed corpus-wide mean segment lengths or discourse-marker frequency tables; the < 0.67 length ratio over Sefaria's segmentation is new arithmetic, sensitive to segmentation conventions, and its threshold value is nowhere in print.

Sources cited by the triage

S. Lieberman, Talmud shel Kisarin (The Talmud of Caesarea), Tarbiz Supplement II.4 (Jerusalem, 1931)
Y. Sussmann, 'Ve-shuv li-Yerushalmi Neziqin', in Y. Sussmann & D. Rosenthal (eds.), Mehqerei Talmud I (Jerusalem: Magnes, 1990)
E. S. Rosenthal & S. Lieberman, Yerushalmi Neziqin, edited from the Escorial manuscript (Jerusalem: Israel Academy of Sciences, 1983)

Predictions

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

Resolution: Killed

Caveats: (a) The QUALITATIVE direction survives: R = 0.8719 < 1 (pooled cross-check 0.8856, same direction), i.e. the Bavot ARE terser than the rest of the Yerushalmi, consistent with Lieberman's 'Talmud of Caesarea' thesis (Talmud shel Kisarin, 1931; cf. Sussmann 1990) that Y. Nezikin is a separate, terser redaction. What fails is the specific magnitude the conjecture pinned: a mean segment length below two-thirds of the rest (R < 0.67); the observed 487.0 vs 558.6 chars is a ~13% deficit, not the wagered >33% one. (b) The secondary discourse-marker clause also fails as pinned: the conjecture predicted at least one high-frequency redactional marker differing by >=2x in per-segment rate between the Bavot and the rest; the measured marker (Hebrew tani, the baraita-citation formula) runs 0.537/seg in the Bavot vs 0.727/seg in the rest, ratio 0.739 - a ~1.35x difference in the OTHER direction (the Bavot use the formula LESS often, not >=2x more). This clause is non-verdict by registration; recorded for the record only. (c) Pattern note: this kill has the same shape as the PELLA kill and other recent kills - the conjecture's direction is right but its bold quantitative threshold overshot the data; what dies is the pinned magnitude, not the underlying philological thesis, which the measurement in fact mildly corroborates. (d) Text provenance: Sefaria served the Guggenheimer critical edition (De Gruyter, 1999-2015) for 38 of 39 tractates; Niddah alone came back as a merged composite across its three available witnesses - immaterial to the aggregate (a single small non-Bavot tractate; the Bavot group is uniformly Guggenheimer, and the unweighted and pooled averaging methods agree).

Registered against Sefaria's open Talmud Yerushalmi (verified live open API/export). PREDICTION VERBATIM: Prediction: measuring mean segment (sugya-unit) character length across Sefaria's machine-readable Talmud Yerushalmi, the three Nezikin tractates Bava Kamma, Bava Metzia and Bava Batra will show a mean segment length below two-thirds (a ratio < 0.67) of the mean across the other extant tractates, and at least one high-frequency redactional discourse marker will differ in per-segment rate by a factor of two or more between Yerushalmi Nezikin and the rest - primary clause: the < 0.67 Nezikin-to-rest mean-segment-length ratio; the verdict follows it. The test voids for coverage if the Bavot segmentation cannot be aligned to the rest of the corpus.

Resolution criteria — the registered fine print

Resolution criteria: DENOMINATOR (verbatim): mean segment (sugya-unit) character length across Sefaria's machine-readable Talmud Yerushalmi, per tractate. NUMERATOR GROUPING: the three Nezikin tractates Bava Kamma, Bava Metzia, Bava Batra vs the mean across the OTHER extant Yerushalmi tractates. DATA: fetch each Yerushalmi tractate from the Sefaria API/export; segment = Sefaria's own smallest addressable text unit (the halakhah/segment level); mean segment char length = total Hebrew chars / segment count, per tractate; freeze snapshot+sha256. R = mean(Bavot three) / mean(all other tractates). SECONDARY (non-verdict): at least one high-frequency redactional discourse marker differing by >=2x per-segment rate between Bavot and rest. CLAUSE PRECEDENCE: (1) INCONCLUSIVE_BY_DESIGN (the conjecture's void rule) if the Bavot segmentation cannot be aligned to the rest of the corpus (e.g. structural incompatibility); (2) SUPPORTED if R < 0.67; (3) REFUTED if R >= 0.67. Compute firewalled from threshold and direction.

Known-priors disclosure — what the registrant already knew

Known priors disclosure: Triage (2026-07-17) graded ADJACENT: Lieberman's 'Talmud of Caesarea' thesis (Y. Nezikin a separate, terser redaction) is textbook qualitatively (Lieberman 1931; Sussmann 1990) but no corpus-wide mean-segment-length computation over Sefaria exists. The segment-alignment void is a live risk. Direction genuinely open.

Method and dataset — how it was measured

Register-before-compute: prediction registered verbatim in conjecture_prediction_batch4_20260718.json (file timestamp 2026-07-18T12:15:13Z) before any measurement; compute firewalled (claude-sonnet ran the measurement blind to the 0.67 threshold and to the wagered direction), computed_at_utc 2026-07-18T12:24:38Z with API fetch window 12:20:42-12:21:28Z, postdating registration. Segment definition: Sefaria's smallest addressable Yerushalmi unit, the depth-3 leaf (Chapter>Halakhah>Segment) of each tractate's Hebrew array, HTML-stripped and whitespace-normalized; every one of the 39 extant tractates was confirmed per-response to share this identical JaggedArrayNode depth-3 schema, so the registered coverage-void clause (Bavot segmentation unalignable to the rest) is NOT triggered. R = unweighted mean of per-tractate mean segment char lengths over the three Bavot divided by the same over the other 36 tractates; pooled/volume-weighted ratio computed as cross-check (0.8856, same direction, within 0.014). Registered clause precedence applied: (1) void not triggered; (2) SUPPORTED requires R < 0.67; (3) observed R = 0.8719 >= 0.67, therefore REFUTED.

Dataset: Sefaria Yerushalmi, 39 tractates (Guggenheimer critical ed.), depth-3 segment; bavot_mean 487.0 vs rest_mean 558.6 chars

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.