The digest that let the orders die

Claim (verbatim)

Isaac Alfasi's Sefer ha-Halakhot (the Rif, eleventh century) is not a commentary but a digest: it walks the Babylonian Talmud and extracts the operative halakhah, silently dropping the dialectic, the aggadah and - decisively - the entire body of law with no post-Temple application. The Rif omits the orders Kodashim and Tohorot almost in their entirety, and most of Zeraim, because a Jew in exile does not offer sacrifices, guard purities, or (outside the Land) tithe. The mechanism is supersession by selection: once the Rif became the through-studied halakhic spine - and the Rosh and the Tur built directly on it - the practical curriculum aligned to what the Rif kept, and the material he cut lost the recopying demand that keeps a text alive. The digest thus both records and enacts a triage. Prediction restated: measured against the Bavli it digests, the Rif's coverage collapses to the practically-applicable tractates - near-total for the used parts of Moed, Nashim and Nezikin, near-zero for Kodashim, Tohorot and non-Berakhot Zeraim - so the Rif reproduces well under half of the Bavli's daf-extent, and the omitted orders map precisely onto the same tractates that lost their independent base commentaries.

Prediction clause (verbatim)

Prediction: comparing the Rif's coverage to the Babylonian Talmud tractate by tractate in Sefaria, the Rif will carry text for under 45% of the Bavli's daf-extent (a coverage share < 0.45), with coverage near zero (< 0.1) for the orders Kodashim and Tohorot and for Zeraim excluding Berakhot, and near-complete for the practically-applied tractates of Moed, Nashim and Nezikin - primary clause: the < 0.45 overall Rif-to-Bavli daf-coverage share; the verdict follows it. The test voids for coverage if Sefaria's Rif is not mapped to Bavli daf references.

Kill-dataset (verbatim)

Kill: Sefaria's open Rif (Hilchot HaRif / Sefer ha-Halachot) keyed against the Babylonian Talmud daf structure - compute the fraction of Bavli daf-extent the Rif reproduces, per tractate and per order, with the Vilna Rif pagination as the control.

Novelty / leakage triage

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

At tractate level the omission map is textbook - every survey states that the Rif covers Berakhot, Moed, Nashim, Nezikin, Hullin and the Halakhot Ketannot and drops the rest of Kodashim, Tohorot and Zeraim as non-applicable - so that half of the claim is already in print (Elon; Ta-Shma). But the primary clause is the overall daf-extent fraction (< 0.45), which turns on the within-tractate daf-presence of a digest that skips dialectic and aggadah unevenly; no daf-coverage tabulation of the Rif exists, and honest estimates straddle the threshold (covered tractates hold roughly 0.6-0.67 of Bavli daf-extent, times an unmeasured within-tractate presence). Genuine new tabulation with an uncertain outcome; note also that the Rif's full Hullin makes the packet's < 0.1 Kodashim sub-clause doubtful.

Predictions