# v2.1 stable implementation decision record

Model: `2.1.0`  
Profile: `fixed-five-event-response-v2.1`  
Score profile: `fba67e0308b51c8c28526456a327e6e31638852f7a73b492e856dda3a7546b11`

## Decision

Promote the unchanged v2.1 scoring construct to a stable implementation release for continued external review, **not** as a validated power-system security metric.

## Score-driving catalogue

- `NORDIC_REFERENCE_INCIDENT_CASE`
- `HYDRO_UNIT_LOSS_CASE`
- `FINLAND_IMPORT_DEPENDENCY_STRESS`
- `WIND_COMMON_MODE_CASE`
- `LARGE_LOAD_DISCONNECTION_CASE`

## Selected construct

Scenario response is event-conditioned initial RoCoF. The normalized Nordic inertia curve is system-state context only. Missing fixed-core input suppresses the point score. JAO CNE/PTC and physical-link observations remain separate context.

## Rejected constructs

- v2.0 min(inertia_context,event_response) per scenario because P->0 was discontinuous
- seven-case live/five-case historical scalar because catalogue definitions differed
- generic HVDC import/export as independent score drivers
- DK2-DE area-pair labelled as a pure Kontek trip
- cross-domain scalar combining frequency with JAO constraints
- negative published slack labelled as material reconciliation failure