Evidence boundary
Promising for efficiency, not delivery.
The pilot was deliberately small and descriptive. Generation 2 is the common perturbation, never a policy result.
01 · Causal response
Protection factor by generation
Select a traffic class, then step through the generation sequence. Generation 3 is the first plan made after the terminal failure.
selected-class factor
selected-class factor
selected-class factor
02 · Post-shock cost
Initial protection + emitted repair
Five prescheduled generations, 200 source symbols per lifecycle. Requested repair is shown as context but is not added to the wire total.
03 · Whole-run transport
Same binary, two conditions
Means across the two randomized blocks. Change the outcome without changing the treatment source.
04 · Delivery guardrail
Delivery saturated after the shock
Both conditions are shown in frozen block order. In regime-change-v1 the shared generation-2 failure explains 7/8; generations 3–7 were 5/5 for every policy.
| Condition | Policy | Block 1 | Block 2 | Interpretation |
|---|
05 · Provenance
Trace every view to frozen evidence
The explorer is a generated reporting surface. It performs no network request and contains the evidence input inline.