Congress Monitor Build Handoff

Area: EO Capture Score Methodology (eo_capture_scoring) Filed: 2026-04-20T05:50:43.999Z Source: External LLM via /handoff/congress

Briefing Sent

Refine the formula that turns donor concentration + voting-against-constituent + silence + contradiction signals into the 0-100 capture score visible on /congress. The output is a methodology spec + a recommended weight vector + 3-5 worked examples.

Result

Goblin House Capture Score Methodology v1.2

The Capture Score is a deterministic metric representing the degree to which an Elected Official’s (EO) legislative behavior, public stance, and financial dependencies align with private interests over constituent needs. It is computed as a weighted aggregate of four signal-specific subscores.

1. Input Signals & Source Requirements

All inputs must be grounded in the database as either facts or connections with the appropriate confidence level.

2. Subscore Formulas (0-100)

A. Donor Concentration Subscore ($S_{dc}$)

Measures the "Capture Floor" based on the dominance of the Top 3 Donor Sectors. $$S_{dc} = \min\left(100, \frac{\text{Sum of Top 3 Sectors}}{\text{Total Individual + PAC Receipts}} \times 200\right)$$ Note: A 50% concentration in just three sectors triggers a 100 subscore.

B. Voting vs. Constituent Subscore ($S_{vac}$)

Measures the frequency of votes identified as against_constituent in the voting_records table. $$S_{vac} = \frac{\text{Count(Votes where category = 'against_constituent')}}{\text{Total Analyzed Votes}} \times 100$$

C. Silence/Contradiction Friction ($S_{fric}$)

Combines silences and contradictions (CX), weighted by severity. $$S_{fric} = \min\left(100, (\text{Count}(CS) \times 10) + (\text{Count}(CX_{high}) \times 25) + (\text{Count}(CX_{med}) \times 15)\right)$$

3. Weight Vector & Tier Cutoffs

The final score ($CS_{final}$) is a weighted sum of the subscores:

• Donor Concentration ($W_{dc}$): 0.35
• Voting Alignment ($W_{vac}$): 0.35
• Silence/Contradiction Friction ($W_{fric}$): 0.30

Tier Cutoffs:

• 0-20: Representative. Highly aligned with district/state baseline; diversified funding.
• 21-50: At Risk. High concentration of industry funding; sporadic constituent friction.
• 51-75: Captured. Significant donor-aligned voting; recurring silences on regulatory oversight.
• 76-100: Institutionally Captured. Primary legislative activity serves top donors; routine contradictions of platform.

4. Worked Examples (Derivation)

Example 1: Sen. Ted Cruz

• DC: Top 3 Sectors (FIRE, Oil & Gas, Leadership PACs) = $12.4M of $21.5M (57.6%). $S_{dc} = 100$.
• VAC: 4 'against_constituent' votes out of 10 analyzed (Border Bill, Infrastructure, 2x Healthcare). $S_{vac} = 40$.
• Fric: 1 Silence (SEC 13f-2) + 0 CX. $1 \times 10 = 10$. $S_{fric} = 10$.
• Calculation: $(100 \times 0.35) + (40 \times 0.35) + (10 \times 0.30) = 35 + 14 + 3 = \mathbf{52}$
• Result: Captured

Example 2: Speaker Mike Johnson

• DC: Top 3 Sectors (Leadership PACs, Oil & Gas, Finance) = $3.1M of $7.8M (39.7%). $S_{dc} = 79.4$.
• VAC: 1 'against_constituent' vote (Omnibus) out of 5 analyzed. $S_{vac} = 20$.
• Fric: 1 Silence (Methane Fees) + 1 High CX (Omnibus Reversal). $(1 \times 10) + (1 \times 25) = 35$. $S_{fric} = 35$.
• Calculation: $(79.4 \times 0.35) + (20 \times 0.35) + (35 \times 0.30) = 27.79 + 7 + 10.5 = \mathbf{45.29}$
• Result: At Risk

Example 3: Leader Hakeem Jeffries

• DC: Top 3 Sectors (Securities/Investments, Law, Real Estate) = $4.2M of $15.1M (27.8%). $S_{dc} = 55.6$.
• VAC: 0 'against_constituent' votes identified. $S_{vac} = 0$.
• Fric: 1 Silence (FTT Revenue) + 1 Med CX (DHS Whipping). $(1 \times 10) + (1 \times 15) = 25$. $S_{fric} = 25$.
• Calculation: $(55.6 \times 0.35) + (0 \times 0.35) + (25 \times 0.30) = 19.46 + 0 + 7.5 = \mathbf{26.96}$
• Result: At Risk

5. Known Limitations & Falsification

• Lagging Indicators: FEC data is updated quarterly. A recent capture event (e.g., a massive dark money influx) may take 90 days to impact the score.
• Small Sample Sizes: EOs with fewer than 5 analyzed votes default to an $S_{vac}$ of 0, potentially under-representing capture.
• Falsification: A score is falsified if (a) an underlying roll call vote is retracted or misattributed, or (b) a donor connection is proven to be an individual donation miscategorized as a PAC sector contribution by OpenSecrets.

6. Implementation Notes for capture-score.ts

/**
 * REFINEMENT: Ensure all inputs are escHtml() in the view.
 * Logic uses Drizzle to aggregate connections for dc calculation.
 */
export async function calculateCaptureScore(entityId: string): Promise<number> {
  const dc = await getDonorConcentration(entityId); // DC Subscore
  const vac = await getVotingFriction(entityId);     // VAC Subscore
  const fric = await getStanceFriction(entityId);    // CS + CX Subscore

  const finalScore = (dc * 0.35) + (vac * 0.35) + (fric * 0.30);
  return Math.round(finalScore);
}