# L3 — Rate feasibility & validation

**NO E3 score.** Ontology / measurement feasibility only.

## Monthly / quarterly event rates

| Cadence | Narrow federal party_switching | Expanded ontology (types 1–8) | Notes |
|---------|--------------------------------|-------------------------------|-------|
| Monthly | **INSUFFICIENT** density | Still **weak**; many structural zeros; campaign seasonality | Not honest as continuous velocity sensor |
| Quarterly | **Poor–marginal** | **Marginal** if endorsements/repudiations included | Prefer rare-event / zero-inflated framing |
| Annual / Congress | **Feasible** for SF1/SF2 backbone | Feasible | Too coarse for “velocity early warning” claims |

**Feasibility conclusion:** A monthly E3 velocity series for US federal elites is **not honest** with currently available structured data if restricted to true defections. Quarterly only as experimental with expanded ontology + heavy FP controls. Congress/annual structured switching rates are feasible but answer a coarser question.

## Structured-data availability summary

- **Strong:** VoteView party codes; rollcall matrix; Congress.gov membership dates.  
- **Medium:** Academic switcher rosters; Ballotpedia modern tables; Congressional Record procedural acts.  
- **Weak:** Protest-rationale resignations; long-run endorsement panels; prosecutor/judicial protest panels.

## Validation gate (before any future score conversation)

- [ ] Elite universe frozen  
- [ ] Event type dictionary frozen with includes/excludes  
- [ ] Dual coding IRR reported  
- [ ] Forbidden substitutes list enforced in pipeline tests  
- [ ] Exposure denominator defined  
- [ ] No outcome/transition tuning  

## Relation to other lanes

- L1 rigidity / E2 VoteView-CBS = **LEVEL / permeability** sensors — complementary, not E3.  
- Do **not** reopen exploratory S1/E2 scoring work for this lane.