Our Methodology
At Karmameter, we believe that public trust must be earned through rigorous, verifiable data. Our methodology is built on a commitment to absolute objectivity, transparency, and data integrity. We do not editorialize; we quantify.
1. Data Acquisition
We aggregate raw data from over 12,000 unique municipal and federal sources. Our automated scrapers run daily cycles to ingest records from disparate formats, ensuring real-time accountability.
Ingestion Sources
2. Verification Process
Raw data is often messy or incomplete. We employ a multi-stage verification pipeline that combines algorithmic anomaly detection with human-in-the-loop auditing for high-stakes records.
def verify_record(record):
# Cross-reference with secondary source
if record.source_id not in trusted_sources:
flag_for_human_review(record)
return "PENDING_REVIEW"
# Check for statistical anomalies in spending
if record.amount > (avg_spend * 3.0):
record.anomaly_score = HIGH
- Entity Resolution: Merging duplicate profiles (e.g. "J. Doe" vs "John Doe").
- Anomaly Detection: Flagging budget outliers that deviate by >2 standard deviations.
- Human Audit: A team of data analysts manually reviews all flagged items before publication.
3. Scoring Framework
The "Karma Score" is a normalized index from 0 to 100 representing an official's adherence to fiscal responsibility and transparency mandates. It is calculated via a weighted average of three core pillars.
Fiscal Efficiency
Budget utilization vs. project completion rates.
Transparency
Timeliness of disclosures and public record availability.
Presence
Attendance in council meetings and voting records.
4. Peer Review & Feedback
We recognize that algorithms can contain implicit biases. Our scoring logic is documented and available for structured review by qualified researchers and policy experts. We regularly invite external feedback to refine our weighting mechanisms and address potential blind spots.
Suggest a Modification
If you believe a specific metric unfairly penalizes rural municipalities or has other systemic issues, please reach out via our contact form or contact our Data Ethics Board.