# L4 ↔ E2-N4 overlap (preserve denominator audit; no composite)

**Job:** `lane-L4-boundary-permeability`  
**Canonical N4 pack:** `docs/e2/WAVE2C_E2_N4_OFFER_ATTRACT_SERIES.md` · `WAVE2C_E2_N4_COSPONSOR_FEASIBILITY.md`  
**Audit JSON:** `data/e2/n4_definition_audit.json`

## 1. What overlaps

| L4 ID | E2-N4 object | Overlap type | L4 action |
|-------|--------------|--------------|-----------|
| L4-A1 offer | `cbs_offer_out_party_original_cosponsorship_rate` | **Identity reuse** | KEEP; cite landed series; do not fork definitions |
| L4-A2 attract | `attract_aggregate_unweighted` chamber mean | **Identity reuse** | KEEP separate; forbid average with offer |
| L4-D1 reciprocity gap | attract−offer descriptive | **Derived display only** | Allowed as audit visualization; **not** a score input |
| L4-A3 Congress.gov rebuild | lineage A | **Replication of same construct** | MODIFY until API key + subset match |
| L4-B1 edge networks | CBS/Congress micro cosponsor edges | **Shared microdata, different aggregation** | MODIFY; member-mean ≠ edge-mean (audit already warns) |

Non-overlap (intentionally separate): L4-C* mass social, L4-E* civic bridging, L4-X* polarization/cohesion/affect guardrails.

## 2. Denominator / definition locks (do not reopen)

From `n4_definition_audit.json` — **preserve**:

1. **Offer** = share of member-term original cosponsorships on out-party-sponsored bills (cosponsor date = introduction date).  
2. **Attract** = share of original cosponsors on own bills who are out-party.  
3. Aggregation primary = chamber×Congress **unweighted member-mean**.  
4. Zero original-activity members excluded from that series’ mean (`n_offer`/`n_attract` may be < `n_members`).  
5. Party-opportunity reweight **not** applied.  
6. Unit = cooperation proportion ∈ [0,1] — **no inversion** this pack.  
7. **Rule:** Do NOT generalize to “all cooperation declined.” Offer↓ and attract↑/flat are contrary — preserve both.

## 3. What L4 must NOT do to N4

- Build `0.5*(offer+attract)` or any unnamed composite.  
- Invert offer into a “fragmentation score” without a separate named memo.  
- Replace N1/N2 contrary evidence with N4.  
- Claim N4 validates full E2 or L1 Rigidity Index.  
- Recompute chamber means from a different denominator quietly.

## 4. Endpoint reminder

CEL CBS coverage in tree: Congress **98–118**. Primary completed endpoint for descriptive Δ = **118**. No Congress 119 claim in this lane fill.

## Explicit non-claims
`score_authorized(E2)=0`. L4 does not authorize E2 scoring.