Some judges want a letter or conference before you file a motion. The requirement sits in a standing order or courtroom procedure and can change by judge and motion type. We counted where the database shows that step.
The unit and the population
The unit is a judge-court entry, v2_judges.id within its district_id listing. Count each entry once, however many matching rules or versions it has. A person in several courts can appear repeatedly. A linked source can be an inherited order, an adopted court policy, a form or a docket copy of standing procedures. Repeated text or a roster association alone leaves independent adoption unproven.
The active roster holds 574 federal entries in 19 covered district listings. Of those, 516 have their own rule-bearing documents. Southern District of New York has 65 roster entries and 64 own-document entries. Eastern District of New York has 47 of each. We read production on October 22, 2026, Pacific time. Paginated GETs do not give a consistent snapshot. No write, RPC or DDL ran. We excluded the statewide California rulebook and the duplicate redirect stubs; no stub counts as another covered listing.
The fields we searched
The structured type is content.structured_data.logic_type = PreMotionConferenceRule, with fields pre_motion_letter.required, trigger, exemptions and sequential_requirements[] (step, with, required, must_complete_before). Missing booleans are unknown. Extraction flags can misdescribe counsel-only meetings or optional procedures. Operative content.source_text supplies evidence, and content.citation_source gives a locator when available.
A candidate union takes the structured-type predicate and a case-insensitive source-text search. In build.py, the regex searches pre-motion and pre-filing terms and letters or conferences near motion. Searches read stored extractions, so original PDFs go unread. Joins run court to judge, source to judge entry, document to source and rule to document. v_rule_sources_current filters active sources and identifies the latest active version by fetch or create order. Candidates stay only on that version and within the own-document population.
The review step
A match needs a reviewed source-text passage that establishes a required court-facing letter, conference or request for one before filing or making a substantive motion, for at least one specified kind of motion or case. A court-facing email or written conference-request form counts as a letter or request. A request the court can deny or waive counts when the request itself is required. Model case-management and scheduling orders count even when dates stay blank. Exceptions, consent-case limits and motion-specific conditions still apply. We never expand an omitted scope to every motion. Explicit exceptions to an optional procedure count when they name a category requiring it, such as discovery motions with formal briefing. An incorporated local rule with an explicit pre-filing instruction can establish a match; a cross-reference alone cannot. A procedural motion requesting a conference is the initiating request and sits outside the substantive motion the requirement governs.
The screen leaves out counsel-only meet-and-confer duties, optional or encouraged conferences, courtesy requests, scheduling or settlement conferences, hearing-date reservations, ordinary briefing schedules and format or response requirements that do not establish the initiating obligation. It also leaves out conferences required only after a substantive motion has been filed or before its hearing. A generic statement that a court may impose a conference requirement is insufficient. Ambiguous fragments and links or headings alone do not establish a match. Required letters used solely as the substantive discovery motion itself, without evidence of a preceding step, do not meet this definition.
Review decisions sit by full passage hash in internal/reviews.json, and the corpus fingerprint must match the reviewed snapshot or the builder stops. Other hits do not become matches by default.
What the numbers mean
Reviewed text supplies a match for 132 of the 516 federal entries with their own documents, or 25.6 percent. In the Southern District of New York, 45 of 64 entries match, or 70.3 percent. In the Eastern District of New York, 40 of 47 entries match, or 85.1 percent. These are conservative evidence counts from stored extracted text. An unmatched entry stays unconfirmed, and the judge may still have such a rule. Search and screening can miss language or leave ambiguous wording unconfirmed. The counts include conditional procedures and policies carried on a judge's page. They do not measure independent adoption or requirements applying to every motion.
The final source round completed 10 comparisons after 14 random draws. It found no classification mismatch and one extraction typo, corrected locally. Four current direct sources stayed unverified and were replaced for the completed sample. Guzman's source comparison contradicted the stored pre-filing requirement, so her entry is excluded locally. Bulsara's current wording and Wicks's extraction typo have local quote corrections. The database keeps the chosen extraction in database_evidence_quote, while evidence_quote can carry a marked source correction. The per-court counts are at https://www.courtrules.app/data/pre-motion-conferences
Top comments (0)