The lect layer goes load-bearing — 408,884 stage assignments

4 August 2026 · Nabu news

A week ago the nabu-lects registry gave the library a vocabulary for historical language stages — lat:med, akk:ob, grc:koi — and a read seam that resolved codes through it. This phase makes the vocabulary load-bearing: 408,884 journaled stage assignments (4 August 2026), derived from evidence the catalog already held and stored in a journal that survives rebuilds, rides backups, and states its basis on every row.

Three grains of evidence, in the order they were ruled:

The registry itself grew to meet the corpus: Sumerian minted its five-stage ladder, proto-cuneiform (qpc) became its own anchor with the field’s genuinely open language-identity question stated in its note rather than decided, and Akkadian gained the Assyriological six-way Old/Middle/Neo × Assyrian/Babylonian grid.

Honesty machinery shipped alongside: nabu lect check-dates audits every assignment against its document’s own dates. On the very first live run it caught one — an inscription whose extracted dating interval ran backwards had slipped through containment — and the inference now refuses reversed intervals outright. The resolved stage is an ordinary document facet, so show prints it, search --lect filters by it at any grain, and every nabu language card ends with a live stage ladder.

← All news  ·  Atom feed