Constitutional Fidelity Index

An interactive dashboard that evaluates executive orders across seven constitutional dimensions using six distinct interpretive legal lenses. Every score is computed from a transparent, multi-step aggregation pipeline — no black boxes.

100 executive orders evaluated across four administrations (Obama, Trump I, Biden, Trump II).

---

What It Does

Each executive order is scored on a -2 to +2 scale across seven constitutional dimensions, then evaluated independently through six legal interpretive lenses. The aggregation engine produces:

• A CFI Score (0–100) representing overall constitutional alignment
• A Constitutional Floor indicator: CLEAR, CAUTION, or VIOLATION
• A Steelman Delta — how much the strongest-case defense diverges from the consensus evaluation
• A Contested Dimensions list — where the lenses most disagree

The radar chart renders the mean, min, and max scoring envelope across evaluative lenses, making interpretive disagreement visible at a glance.

---

Seven Constitutional Dimensions

Code	Dimension	Primary Sources
R	Individual Rights & Liberties	Amendments 1–4, 9, 13, 14
E	Equal Protection & Non-Discrimination	14th Amendment, Civil Rights Act
D	Democratic Legitimacy & Representation	Articles I–II, Amendments 15, 17, 19, 26
SP	Separation of Powers	Articles I–III, Federalist 47–51
DP	Due Process & Rule of Law	Amendments 5–8, 14
W	General Welfare & Common Good	Preamble, Article I §8
N	National Sovereignty & Security	Preamble, Article II, Article IV §4

---

Six Legal Lenses

Lens	Method	Source Access
Textualist	Plain-language reading of constitutional text	Constitutional text only
Originalist	Original public meaning at ratification	Text + Federalist Papers, Convention records
Doctrinalist	Established judicial doctrine and stare decisis	Text + legislation + Supreme Court precedent
Living Constitutionalist	Constitutional development arc and trajectories	Text + founding + precedent
Pragmatist	Real-world consequences in comparable jurisdictions	All sources + V-Dem, OECD, empirical research
Steelman	Strongest genuine constitutional defense	All sources, good-faith synthesis

The Steelman lens does not factor into the CFI Score — it is measured separately as a delta, showing how much the best-case argument moves the needle.

---

Aggregation Algorithm

The CFI Score is not an average. It uses a relevance-weighted aggregation across evaluative lenses:

1. Relevance Filtering — dimensions with mean relevance < 0.2 are excluded
2. Constitutional Floor — if 3+ evaluative lenses score −2 on any dimension, the action is flagged VIOLATION
3. Alignment Score — relevance-weighted mean across relevant dimensions, normalized to 0–100
4. Steelman Delta — signed difference between Steelman and evaluative-lens consensus
5. Contested Dimensions — dimensions with cross-lens variance > 1.0

Full algorithm: engine/aggregation.js | Schema: engine/schema.json

---

System Prompts

The prompts/ directory contains seven production-quality system prompts (2,791 lines total) for running the evaluation pipeline through any LLM:

prompts/
  00-classifier.md          # Classifies action type, scope, targeting, reversibility
  01-textualist.md          # Plain-text constitutional analysis
  02-originalist.md         # Original public meaning analysis
  03-doctrinalist.md        # Stare decisis and established tests
  04-living-constitutionalist.md
  05-pragmatist.md          # Empirical, comparative analysis
  06-steelman.md            # Best-case defense

These prompts are self-contained and ready for direct injection as LLM system prompts.

---

Tech Stack

Layer	Technology
Framework	React 19
Build	Vite 7
Language	TypeScript / JSX
Charts	Custom SVG (no chart library)
Styling	Inline styles with design tokens
Deployment	Static (Vite build → dist/)

Redesign (in redesign/): Next.js 15, TypeScript, Tailwind v4 — work in progress.

---

Running Locally

# Install dependencies
npm install

# Start dev server
npm run dev
# → http://localhost:5173

# Build for production
npm run build

# Preview production build
npm run preview

The redesign:

cd redesign
npm install
npm run dev
# → http://localhost:3000

---

Scoring Reference

Score	Meaning
+2	Action clearly strengthens / directly supported by text or precedent
+1	Action complies; reasonable interpretation
0	Neutral; text silent or arguments balanced
−1	Action conflicts with; weak support or contested
−2	Clear violation of text, doctrine, or precedent

CFI Range	Label
81–100	Strong Constitutional Alignment
61–80	Moderate Constitutional Alignment
41–60	Mixed Constitutional Alignment
21–40	Significant Constitutional Tension
0–20	Severe Constitutional Tension

---

Repository Structure

cfi/
├── src/
│   ├── App.jsx               # Main dashboard — all pages and components
│   ├── main.jsx              # React entry point
│   └── data/
│       └── executiveOrders.js  # 100 evaluated EOs with full scoring data
├── engine/
│   ├── aggregation.js        # 6-step CFI aggregation pipeline
│   ├── schema.json           # JSON Schema for evaluation output
│   └── IMPLEMENTATION_GUIDE.md
├── prompts/                  # 7 LLM system prompts for live evaluation
├── integrity/                # Framing symmetry tests and cross-lens pairs
├── docs/                     # Planning documents
├── redesign/                 # Next.js redesign (WIP)
├── CFI_Whitepaper_v4.pdf     # Full methodology whitepaper
└── CFI_Policy_Brief_1.pdf    # Summary policy brief