# WAVE2C E2 — Elite Fragmentation narrow sensor research

**Agent (proposer):** grok · **Reviewer (required):** astra  
**Job:** `b73c292a` · **Research queue:** `9e319e25`  
**Status:** RESEARCH ONLY — **no production scoring** · **Date:** 2026-09-09 ~10:00 AM PT  
**L1:** `L1-US-v0.1` frozen/unchanged · **`score_authorized(E2)=0`**  
**Sequencing:** Astra `3fea3a0c` — after M1-LONG feasibility, **S1 + E2 in parallel** (this memo = E2 leg).  
**Governance:** `decision_authority=delegated_agent_governance` · `delegated_by=owner` · independent review required · **no self-approval**.  
**Hub:** do **not** post until Astra review; leave Hub mirror to parent.

---

## 0. Purpose

Scaffold a **narrow**, research-only sensor pack for construct **E2 Elite Fragmentation**: freeze-aware definition, candidate **public** elite-level sensors that are **not** FSI Factionalized Elites, measurement options with hard role tags, falsification/risks, explicit non-goals, and next evidence pulls.

**Narrow** means: prefer 1–2 auditable elite cohesion/fragmentation series with clear units and provenance — not a multi-source mashup score, and not a rebadge of FSI.

This memo does **not** authorize E2 scores, change L1/M1, or compute Phase/Load/PPS.

---

## 1. Construct definition (frozen methodology)

| Field | Frozen value | Source |
|-------|--------------|--------|
| **ID** | `E2` | `docs/MODEL_SPEC.md`, `srp/constants.py` |
| **Name** | **Elite Fragmentation** | concept_model_version `0.1` (do not rename/replace) |
| **Domain** | Elite (`E`) | |
| **Time scale** | `INTERMEDIATE` | |
| **Coupling** | Member of **C1** Legitimacy × Elite Fracture with L1/L2/L3/E3 | scores UNKNOWN |
| **Authorization** | **`score_authorized=false`**; raw ingest not yet wired as MEASUREMENT | `WAVE1_AUTHORIZATION["E2"]` |
| **Standing FSI rule** | FSI = **CORROBORATION / EXTERNAL_VALIDATION only** — **NOT E2 measurement** | `MEASUREMENT_SPEC.md`, `SENSOR_CANDIDATES.md`, `srp/ingestion/fsi.py` |

### 1.1 Working measurement intent (research gloss — not a formula freeze)

E2 targets **fragmentation among elites** (intra-elite cohesion breakdown: factional splits, leadership challenges, within-party ideological dispersion, elite coalition fracture) — **not** mass affective tribalization (S1), not elite *rule adherence* (E1/E1a), and not *defection velocity* (E3).

Neighbor boundaries (standing + proposed):

| Neighbor | Boundary |
|----------|----------|
| **E1 / E1a** | Rule adherence / V-Dem liberal democracy subcomponent research ≠ fragmentation |
| **E3** | Defection **velocity** (FAST); YoY FSI ≠ E3 |
| **S1** | Mass affective tribalization ≠ elite fragmentation |
| **B2 / ACLED** | Coercion/violence events ≠ elite fragmentation MEASUREMENT |
| **FSI Factionalized Elites** | **Forbidden as E2 MEASUREMENT**; may remain EXTERNAL_VALIDATION only |

---

## 2. Candidate public sensors / sources

### 2.1 Hard gate — FSI (registered but not MEASUREMENT)

| Sensor | Series | Role (frozen standing) |
|--------|--------|------------------------|
| `fsi_total` | FSI total | CORROBORATION only |
| `fsi_factionalized_elites` | Factionalized Elites (FE) | **EXTERNAL_VALIDATION only** — **forbidden as E2 measurement** |

Existing connector: `srp/ingestion/fsi.py` documents the same gate. Do **not** promote FE to MEASUREMENT in this pack.

### 2.2 Narrow MEASUREMENT candidates (public — proposed)

Prefer **U.S. congressional / party-elite** observables with machine-readable history. All mappings remain `approved=false`.

| Sensor (proposed id) | Source | What it measures | Cadence | Role | Why narrow-fit |
|----------------------|--------|------------------|---------|------|----------------|
| `voteview_within_party_nominate_disp` | VoteView / DW-NOMINATE (public) | Within-party dispersion of House/Senate ideal points (e.g. SD or IQR of party caucus on 1st dimension) | Congress / chamber-year | **MEASUREMENT** candidate | Elite ideological fragmentation inside parties — not mass affect |
| `congress_party_unity_invert` | VoteView / CQ-style party unity scores (public aggregates where available) | 100 − party unity (or share of party-line breaks) by chamber-party-year | Annual / Congress | **MEASUREMENT** candidate | Behavioral elite cohesion inverse; document exact vote universe |
| `congress_leadership_challenge` | Biographical / Clerk / news chronology (hand-curated research table) | Count/severity of speakership / conference leadership challenges | Event + annual rollup | CORROBORATION (unless coding protocol frozen later) | Sparse; high face validity for “fragmentation” but coding subjectivity |
| `govtrack_party_loyalty` | GovTrack / Congress.gov derived loyalty stats (if stable public) | Mean loyalty / missed party-line votes | Annual | MEASUREMENT candidate **only if** series continuity verified | Verify license + methodology before ingest |

### 2.3 CORROBORATION / EXTERNAL_VALIDATION (not primary)

| Sensor | Source | Role | Caveat |
|--------|--------|------|--------|
| `fsi_factionalized_elites` | FSI | EXTERNAL_VALIDATION | Standing forbid as MEASUREMENT |
| `fsi_total` | FSI | CORROBORATION | Country-year composite contamination risk |
| V-Dem elite / party indices (non-E1a) | V-Dem | CORROBORATION research | Do not silently reuse E1a `v2x_libdem` as E2 |
| Expert elite surveys (e.g. party cohesion batteries) | Various | CORROBORATION | Access / irregular cadence |

### 2.4 Rejected as E2 MEASUREMENT

| Candidate | Why rejected |
|-----------|--------------|
| FSI Factionalized Elites as score input | Standing **FSI FE ≠ E2 measurement** |
| Mass ANES/Pew affective polarization | That is **S1** research territory |
| ACLED elite-targeted violence counts | Behavioral; B-domain adjacency; not fragmentation cohesion |
| Unemployment / sentiment mashups | Material/narrative contamination |
| YoY Δ FSI | Reserved-forbidden near **E3** (`YoY FSI ≠ E3`) |

---

## 3. Measurement options (narrow)

### Option N1 — Within-party DW-NOMINATE dispersion (preferred research primary)

1. Pull VoteView member-congress ideal points (House + Senate separately).  
2. For each chamber × party × Congress (or year): compute dispersion of 1st-dimension NOMINATE among party members (predeclare SD vs IQR; minimum member n).  
3. Optional: mean of majority-party and minority-party dispersion, or report both series.  
4. **No** cross-walk into 0–4 E2 bands.

**Pros:** Public, long history, elite-only, auditable.  
**Cons:** Ideal-point model assumptions; party switching; Southern realignment era comparability; not pure “factional organization.”

### Option N2 — Party unity inverse (behavioral cohesion)

1. Obtain chamber-party-year party-unity (or party-line agreement) series from a documented public method.  
2. Research series: `fragmentation_proxy = 100 - unity` (or equivalent).  
3. Keep VoteView dispersion and unity as **separate** sensors — correlation diagnostic, not averaged score.

### Option N3 — Leadership-challenge event log (corroboration ribbon)

Hand-coded annual count of major U.S. party leadership challenges / failed speakership ballots. Use only as CORROBORATION unless a separate coding protocol passes review.

### Role discipline

- At most **one** primary MEASUREMENT candidate advanced to an evidence pack at a time (N1 recommended).  
- FSI FE may be plotted alongside for EXTERNAL_VALIDATION with an explicit caption: *“not E2 measurement.”*

---

## 4. Falsification / risks

| Risk | Severity | Falsification / mitigation |
|------|----------|----------------------------|
| **Smuggling FSI FE into MEASUREMENT** | Critical | Unit test / doc gate; connector note already forbids; evidence pack must role-tag FE as EXTERNAL_VALIDATION |
| **Confusing E2 with E1a / V-Dem libdem** | High | Separate sensor_ids; E1a remains RESEARCH_NOT_AUTHORIZED as E1 |
| **Confusing E2 with S1 mass affect** | High | Elite-only member universe; reject ANES FT gaps for E2 |
| **Confusing E2 with E3 velocity** | Medium | Levels/dispersion ≠ YoY velocity; no YoY FSI |
| **Realignment / party-system breaks** | Medium | Segment eras; don’t force single trend claim |
| **Chamber pooling** | Medium | House vs Senate separate primary tables |
| **Tuning to known crises** | Process | Independent evaluation controls (Astra sequencing); no Phase labels |
| **Self-approval / premature Hub** | Process | Reviewer=astra; parent posts only after review |

---

## 5. What NOT to do

- Do **not** authorize E2 scores or freeze formulas.  
- Do **not** treat FSI Factionalized Elites as E2 MEASUREMENT.  
- Do **not** compute Phase / Load / Sync / PPS / Coupling from E2 (or L1 alone).  
- Do **not** change L1-US-v0.1 or M1 production/research scoring.  
- Do **not** mash N1+N2+FSI into a single “E2 index.”  
- Do **not** use ACLED or mass affective series as E2 MEASUREMENT.  
- Do **not** post Hub mirrors until Astra-reviewed.  
- Do **not** touch Hub orchestration.

---

## 6. Proposed next evidence pulls

1. **VoteView member-Congress file** — download public ideal-point members data; compute chamber×party×Congress within-party 1st-dimension dispersion; write `data/raw/voteview/` + provenance JSON (URL, retrieve time PT, sha256).  
2. **Party-unity series audit** — locate one stable public unity/loyalty series; document methodology + license; prototype `100−unity` annual table **without** scoring.  
3. **FSI FE external-validation ribbon** — if local/public FSI rows exist, align year-overlap with N1 dispersion for a labeled EXTERNAL_VALIDATION scatter only (caption: not measurement); do not fit E2 thresholds.

---

## 7. Governance / review block

| Field | Value |
|-------|--------|
| `decision_authority` | `delegated_agent_governance` |
| `delegated_by` | `owner` |
| **proposer** | `grok` |
| **reviewer** | `astra` (required; no self-approval) |
| **decision** | *pending Astra review* — research scaffold only |
| **evidence** | This memo; `MEASUREMENT_SPEC.md`; `SENSOR_CANDIDATES.md`; `srp/ingestion/fsi.py`; sequencing Hub `3fea3a0c`; queue `9e319e25` |
| **model_version** | concept `0.1` / measurement_spec `0.1` — **no freeze bump** |
| **rollback_target** | Delete/ignore this research pack; FSI role tags + L1/M1/E2 authorization unchanged |
| **Hub post** | **Deferred** to parent after review |

---

## 8. Paths

| Artifact | Path |
|----------|------|
| This memo | `docs/e2/WAVE2C_E2_NARROW_SENSOR_RESEARCH.md` |
| Parallel S1 pack | `docs/s1/WAVE2C_S1_MEASUREMENT_RESEARCH.md` |
| FSI connector (corroboration only) | `srp/ingestion/fsi.py` |