# L4 — False-positive / polarization-collapse risks

**Job:** `lane-L4-boundary-permeability` · RESEARCH-ONLY · MODEL_CHANGE=NO

## 1. Polarization-collapse modes (reject)

| Risk ID | Collapse move | Why false for L4 | Mitigation |
|---------|---------------|------------------|------------|
| PC1 | Use E2-N3 between-party distance as “low permeability” | Measures separation of ideal points, not cross-boundary acts | Tag L4-X1 INSUFFICIENT for L4 |
| PC2 | Use party-unity ↑ as permeability ↓ | Within-bloc cohesion ≠ boundary traffic | L4-X2 INSUFFICIENT |
| PC3 | Use out-party FT / affective gap as permeability | Affective hostility is S1-adjacent, not interaction | L4-X3 INSUFFICIENT |
| PC4 | Invert cooperation (`1−offer`) and rename “polarization” | Silent sign-flip conflates constructs | Forbid inversion without named fracture memo |
| PC5 | Average offer+attract then trend-shop | Erases audited contrary directions | **No composite** hard gate |
| PC6 | Treat GSS trust decline as elite boundary closure | Wrong layer (S2 corroboration ≠ elite reciprocity) | L4-D3 INSUFFICIENT |
| PC7 | Treat compulsory committee co-service as permeability | Forced co-presence ≠ chosen cross-party exchange | L4-B3 INSUFFICIENT |

## 2. False-positive interaction modes

| Risk ID | Sensor failure | Example | Mitigation |
|---------|----------------|---------|------------|
| FP1 | Symbolic bipartisanship | Low-cost cosponsorship without consequential cooperation | Keep native cooperation units; don’t claim consequentiality; optional LES pathway as separate MODIFY |
| FP2 | Majority-opportunity artifact | Offer rates move with seat share without opportunity weight | Audit flag: party-opportunity reweight still open; don’t overclaim |
| FP3 | Zero-activity selection | Excluding inactive members changes chamber mean | Preserve `n_offer`/`n_attract` reporting from audit |
| FP4 | Edge vs member-mean swap | High-activity members dominate edge graphs | Keep member-mean primary for A1/A2; edge nets = B1 MODIFY |
| FP5 | Survey desirability | Cross-party friendship over-report | MODIFY + item wording audit before KEEP |
| FP6 | Bridging-program selection | Participants already permeable | L4-E2 INSUFFICIENT for population claims |

## 3. Independence claim (honest bound)

Boundary permeability is **measurable without simply reproducing polarization** only if:

- at least one **KEEP** sensor is an interaction/traffic measure (here: L4-A1/A2/D1 from CBS), **and**  
- polarization/affect/cohesion series are explicitly **INSUFFICIENT-as-L4** (L4-X*), **and**  
- no composite/inversion collapses contrary N4 directions.

This lane fill meets that bound at **candidate** level — **not** as an authorized L4 score.

## Explicit non-claims
No hypothesis tests vs historical transitions. No Phase/Load/PPS. UNKNOWN acceptable.