How the record is built
JUDGELENS earns trust by being checkable. This page documents every data source, every calculation, and every limitation we know about.
Primary source: CourtListener (Free Law Project)
Judge records, opinions, clusters, dockets, and citations are retrieved live from the CourtListener REST API, a free service of the Free Law Project. Every record keeps its external identifier, original URL, and retrieval timestamp. CourtListener's coverage of federal Article III judges and published opinions is strong; state-court coverage varies by jurisdiction.
RECAP and PACER
Federal dockets and filings live in PACER, the official fee-based system of the U.S. Judiciary. The RECAP archive mirrors only the documents users have donated. JUDGELENS shows RECAP availability per case and never implies that all PACER documents are available. We do not bypass PACER access controls or fees.
Official courts & state portals
Official federal directories (uscourts.gov) and the Supreme Court site are used for directory information and links. State courts (Texas, Florida, California, New York first) are integrated link-first: most state systems have no public API, so users are directed to the official portal rather than shown scraped or guessed data.
How analytics are calculated
Every metric is computed only from records retrieved and verified at calculation time. Each metric displays its sample size, timeframe, and coverage note — for example: "Based on N opinions retrieved from CourtListener; this reflects only decisions indexed by the provider, not a complete docket record."
Motion grant/denial rates are calculated only when the underlying documents reliably identify the motion and its outcome. Below the minimum sample threshold, the platform displays "Insufficient verified data to calculate this metric" instead of a number. Historical outcomes never establish bias, favoritism, or misconduct, and JUDGELENS never ranks judges.
Lens AI
Lens AI answers only from the records retrieved for a given question, cites the specific documents it used, and identifies when the record doesn't contain an answer. It never invents citations or statistics, and its output is legal information — not legal advice.
Corrections
Anyone can request a correction at corrections.html. Verified errors are fixed; disputed records are annotated. We distinguish allegations from findings and filings from judicial conclusions in every presentation.