2099
stated · refuted
Source
derived from the event's own title: "Amerique and Neville perform at Palace theater (dancing and singing revue) - clipping 2099"
Why
pm-actor-533: pm-actor-533: Self-evidently a parsing error: the earlier pass read the clipping id '2099' in the title as a year.
Reasoning
The event held NO date claim of any kind; its own title states 2099 (year precision, parsed as "year only"). ⚠ A title in this corpus is a DERIVED artefact — an earlier pass wrote it from the passage text — so this claim is title-derived and is marked as such rather than cited as a source document; the underlying text remains the real evidence and would supersede it. Confidence `probable`, never `certain`, for that reason. Precision is never inflated beyond the title's own wording. CALIBRATION (run v1.1/l1.4/titledate, against the 489 events that already hold an independent preferred date): day 96.1%, month 94.1% agreement. Titles naming a REFERENCE event rather than the event itself ("before", "after", "during the week of", "per ... lecture") are refused outright and never reach this write — reading an anchor as the date is the same laundering phase3b committed when it wrote telling dates onto events. || v1.1/l1.4/titledate-correct: REFUTED — written in error by my own run v1.1/l1.4/titledate. the number in this title is a CLIPPING/SCAN IDENTIFIER, not a year — the "Amerique and Neville perform" family carries one, and in several the word "clipping" immediately precedes it. Neville danced with Amerique in the late 1920s-30s, so every year my run took from this family is wrong, including the ones that happen to look plausible. This event held no date before that run, so refuting restores its true prior state (undated) rather than leaving a wrong date on display. The row is kept, not deleted, as the record of what the run did.