The pipeline caught up to the page.

The previous sprint corrected the 15FYP NPC adoption anchor from 15 March 2027 to 12 March 2026 and added quantum information as the seventh live prediction. Both edits landed on the rendered page, but they had only partially propagated into the Python pipeline. The prediction registry was still emitting 21 November 2027 on every row — computed against the old anchor with a uniform eight-month default lag. The quantum row wasn't in the registry at all.

This is the same class of drift the Restriction Index logged when its page and pipeline first reconciled: edits land on the display surface first, the pipeline catches up second, and the gap between them is silent until somebody re-runs the pipeline and overwrites the published claims. This sprint closes that gap on the cascade side. The anchor constant in the pipeline was already corrected in the previous release. The live-predictions tuple is now updated to seven entries with the quantum row added. The registry has been regenerated against the corrected anchor with the per-row lags from the dossier, so resolve-by dates now vary across 3 September 2027, 28 August 2028, 1 March 2028, and 24 February 2029 across the seven 15FYP predictions and the five backtest precedents drawn from the 14th and 13th plans.

A residual one-day mismatch remains between the Python output and the hand-curated page dates — the Python pipeline lands on 3 September 2027 where the page reads 12 September 2027, for example. The cause is a calendar-arithmetic shortcut in the pipeline that treats every month as exactly 30 days. The next release fixes the arithmetic so every resolve-by lands cleanly on the NPC adoption day.

What this opens. The page-to-pipeline propagation pattern generalizes. Every paper in the program now ships with a per-run manifest recording the constants the run used alongside the computed outputs, so future drift between the rendered page and the pipeline output is detectable rather than silent.