When history helps prediction
In finite partially observed stochastic processes, when does recursively updated history recover predictive distinctions unavailable in current observation?
Each package is tracked through its question, scope, protocol and interpretation limits.
The sole D-0016-authorized attempt covered six known RP004 constructions, eight uniform whole-field masks and two query families. All 216 truth rows, 48 projections, 120 pair/mask comparisons, 438 conflicts and 16 certificates independently replay. D-0017 accepts the finite result; RP-006 is CLOSED within that declared corpus and mask class. These classifications reproduce already disclosed examples and do not resolve Q-3 or Q-4.
In finite partially observed stochastic processes, when does recursively updated history recover predictive distinctions unavailable in current observation?
When do local continuation and preservation judgments support a global judgment?
Which distinctions are required for continuation claims under copying, branching and merging?
Can targeted restrictions steer horizon-limited decisions away from states with fewer viable future targets?
Within the declared finite class, which whole fields preserve the full lineage queries, and which preserve the topology/state subset?