Repository path: workshop/translations/hafez-sahar-bolbol/dependence-note.md · rendered 2026-09-09

Contamination measurement — T-hafez-sahar-bolbol-R55-v1

python3 tools/dependence_check.py runs/RS-20260831b-radif-hands/dep_cells.json, raw output at runs/RS-20260831b-radif-hands/dep.json. The lead's 20 rendered lines of غزل ۱۳۰ against the whole of each comparator, not against a matched excerpt — the conservative direction, since a whole-book comparator gives every run in the book its chance.

Both figures are reported, run length and 7-gram count, because run length alone has failed as a proxy three times (CLAUDE.md; notes (bcd); RS-20260731g §4.2). Here they agree: zero shared 7-grams against every comparator, which is the stronger of the two signals and the one S028 showed is normally non-zero even between independent pairs (75 of 78 independent pairs share at least one). Zero against a whole 199-ode volume is unusually clean.

Note (bhb) does not fire. The lead's four earlier Hafez renderings are the ones that produced the project's 37-token self-match record; against this poem the longest self-run is 3 tokens. The poems are different poems, which is the whole of the explanation.

What this licenses and what it does not. It licenses the statement that the rendering was made from the Persian. It does not make the lead an independent third translator of anything: no figure in RS-20260831b-radif-hands depends on the lead's independence from Payne, because the carriage evidence there is coded off Payne's own printed pages.