About, Methodology & Corrections

What this site is

Patent Case Watch is a research starting point: a browsable index of patent-case opinions and orders from the U.S. District Court for the Eastern District of Texas and the U.S. Court of Appeals for the Federal Circuit, with links to the official documents. It is not legal advice, not a complete legal database, and not a substitute for reading the opinions themselves or for citation-checking in a professional research service.

Source of every document

Every record comes from the U.S. Government Publishing Office's public govinfo.gov USCOURTS collection, queried for the two courts with a patent nature-of-suit. Each detail page links to the official govinfo package and PDF and shows the date the archive was retrieved. PDFs are served by govinfo.gov, not re-hosted here.

Coverage limits: govinfo publishes court-transmitted opinions and orders — not complete dockets. Briefs, exhibits, hearing transcripts, and some orders are not in the collection and therefore not on this site. A case's absence here says nothing about whether it exists; the archive reflects govinfo's coverage (documents published from roughly 2005 onward) as of the retrieval date shown on each page.

Update cadence: the archive is re-checked against govinfo for newly published documents from both courts on a rolling basis; new decisions appear at the top of the feed, and every page shows the retrieval date of the current dataset. Note that govinfo itself publishes documents on a delay after the court issues them, so the newest entry here may trail the courts by days.

How records are classified

The bibliographic fields — decision date (from the signature block, "Decided:" line, or filing stamp), judge, and document type — are extracted programmatically from the opinion text and the court's docket-entry line. Federal Circuit Rule 36 summary affirmances are identified from the docket line itself. Documents cross-listed under multiple consolidated appeal numbers appear once, with the other docket numbers noted.

The holding (disposition) and subject-matter tags are determined by an AI language model that reads each opinion together with the court clerk's docket-entry line and selects from a fixed list of dispositions and issue tags — it cannot invent a category. When the disposition genuinely cannot be determined, or the model's confidence is low, the record says "See Opinion" rather than guessing. These classifications are automated and not attorney-reviewed; they can be over- or under-inclusive and are research filters, not legal characterizations — always confirm against the opinion itself.

Most entries display a verbatim excerpt of the document's opening passage, clearly labeled — those are the court's own words, not a summary. A small number of entries carry an editorial summary, labeled as AI-assisted and not attorney-reviewed.

Known limitations

A small number of documents were published by govinfo as scanned images without a text layer; their pages say so and offer the PDF as-is. Where govinfo does not serve a document at all, the record is marked "source document unavailable" and no download is offered. Case names are shown as they appear in court records, lightly normalized for capitalization.

Corrections

If a holding label, date, judge, tag, or link is wrong, please report it — corrections are applied by regenerating the record from the official source document.

A public correction-submission channel has not been configured yet. Until it is, please note the document's govinfo ID (shown in the page address) so the record can be re-verified against the official source.