WHAT WENT WRONG LAST MONTH

GENERATED ON THE FIRST ELIGIBLE PIPELINE RUN OF EACH MONTH, FROM THE APPEND-ONLY LEDGER. DEGRADATIONS ARE PUBLISHED WITH IMPROVEMENTS, AND THE SERIES IS NEVER RESTATED.

THE PLEDGEFeature work suspends when freshness, match rate or slippage degrade past their stated thresholds, until they recover.

01 THIS MONTH

MATCH RATE7.1%Free-text names auto-resolved to a dm+d code this run, of everything the live sources yielded. The rest go to the review queue — never guessed.PRECISION 100.0% · GOLD SET 0

FRESHNESS0.0 hMean hours since the last successful fetch, across scheduled sources. A stale set stamped honestly is fine; a silent one is not.THRESHOLD 24 H

SSP SLIPPAGEn/aMedian SSP expiry revision recorded this month (§7.2). Not resupply-estimate accuracy — see the note below on why.NO THRESHOLD — OBSERVED, NOT PROMISED

TIER ACCURACYn/aA tier records a published act, so there is nothing of Pharmasync's own to be wrong about. This cell stays empty by design.SEE METHODOLOGY §02

02 TWELVE MONTHS, NEVER RESTATED

MONTHMATCH RATEFRESHNESSSSP SLIPPAGEGOLD SET

00:00 · 1 SEP 20267.1%0.0 hn/a0

00:00 · 1 AUG 202692.8%0.0 hn/a0

03 CORRECTIONS

NOT YET TRACKED. NOT SUMMARISED HERE YET.

04 SOURCE FRESHNESS

Coverage extends beyond England — see /coverage.json for the full per-nation breakdown.

SSP · ENG0.0 H AGO

SSP · SCO0.0 H AGO

SSP · WAL0.0 H AGO

SSP · NIR0.0 H AGO

CONCESSIONS · ENG0.0 H AGO

MSN · ENG0.0 H AGO

RESTRICTED_EXPORT · ENG0.0 H AGO

RESTRICTED_EXPORT · SCO0.0 H AGO

RESTRICTED_EXPORT · WAL0.0 H AGO

RESTRICTED_EXPORT · NIR0.0 H AGO

LIMITATIONS

tier_accuracy is not published: a tier records a government act (D-009), so there is nothing of Pharmasync's own to be wrong about — see /methodology.estimate_error_days measures SSP expiry slippage (the same signal /slippage.json publishes), not resupply-estimate accuracy. The design's original metric needs MSN resupply-date extraction and MSN resolution detection, neither of which exists yet — a future sprint, not a stated absence like tier accuracy.corrections is null, not an empty array: Pharmasync does not yet track "what went wrong and how it was fixed" in a machine-readable form.match_rate_precision is only ever published when gold_set_size is zero that month — every auto-match was character-for-character the dm+d name, so there was nothing for a person to grade. A month with a nonzero gold_set_size and no precision figure means real judgement calls occurred and were not labelled; that row stays that way permanently, the ledger is never restated.