Evidence that remembers its reasons
Selective correction preserved independent support until its own evidence was withdrawn.
Inspect experiment, replay and evidence →Research evidence · V0–V5
Experiment catalog
Successive bounded experiments, preserved individually. Results include limitations, supporting records and interactive explanatory replays where available.
V0–V3 original excerpts are withheld pending source-byte verification; qualified historical summaries are available; V4 has a structured comparison (not a verbatim terminal transcript); V5 selected original console lines are withheld pending publication clearance. The diagrams are illustrations, not live tests.
Browse the research record
6 experiments shown
Protected official reference
An unauthorized change to a protected reference was rejected; tested state was unchanged.
Inspect experiment, replay and evidence →Accumulated dependency consequences
All 12 explicitly linked dependents were invalidated; separate valid items survived.
Inspect experiment, replay and evidence →Contextual applicability
Stored knowledge remained valid while use outside its dataset context was out of scope.
Inspect experiment, replay and evidence →Comparing knowledge-revision methods
Conventional dependency tracking and CRSE both invalidated 12/12; no superiority. A weak memory condition retained 14 unjustified conclusions.
Inspect experiment, replay and evidence →Knowing when knowledge does not apply
Dataset v1 UNKNOWN, dataset v2 OUT_OF_SCOPE; supporting evidence UNKNOWN and source provenance retained.
Inspect experiment, replay and evidence →