What the comparison establishes
Six of twelve outputs differ between the fixed information-set reconstruction and the latest fictional revision. One original record is Unavailable and four are Boundary. These are deliberately designed properties of a teaching fixture, not a measured revision rate, prediction score or strategy result.
Every observation keeps its own cutoff. The latest column uses all fixture revisions and is deliberately retrospective. A changed label may reflect a sign change, a newly available value or a withdrawn input; it does not necessarily describe an economic transition.
What the archive does not contain
There were no publication events in the fictional historical period. The actual edition date is 2026-10-16, and all twelve reconstructions were generated together for it. File hashes identify released bytes but cannot establish that a label existed years earlier. This distinction applies equally to the JSON records, CSV exports and chart.
Version register
Edition 2026-10-16 is the first version. Fixture synthetic-v1 and rule threshold-v1 are fixed within it. A later correction must identify the affected records and preserve this edition; no prior public edition or revision history is invented here.
