A record must pass every gate.
- The record exists in the normalized campus-rules data layer and is current rather than superseded or archived.
- The rule, normalized subject, institution, and every narrower scope entity are published.
- Every cited source is official, published, current, and approved by the source-verification gate.
- The current verification event is published and approved, matches the exported wording, status, scope, conditions, exceptions, and academic-year field exactly, and has not expired at the cutoff.
- A verify record may be included only when the approved official-source wording itself requires assignment-level confirmation; it remains non-definitive in every export and analysis.
The scope travels with the claim.
Version 1.0.0 contains 13 rule records from 3 institutions: San Diego State University, University of California, Berkeley, University of California, Los Angeles. A record retains the narrowest institution, campus, hall, room-type, academic-year, and population label supported by its official source.
Representativeness: Descriptive of the included records only. The dataset is not a national, state-wide, or statistically representative sample of colleges or residence halls.
A fixed historical snapshot.
The data cutoff is August 19, 2026. Eligibility and staleness are evaluated as of that date. Later source checks or policy changes belong in a new dataset version and do not rewrite this release.
Official evidence first.
Official university and housing sources are the authority. For this release, each record passed isolated extraction, at least two verifier passes, adjudication, unanimous direct-support and scope checks, a minimum 0.95 confidence threshold, fingerprint matching, and deterministic publication blockers. Automated review is not university endorsement or human expert review.
Silence is not a policy.
Missing coverage stays missing. The dataset does not infer that an unrecorded institution allows, prohibits, or provides an item. Analyses state whether missing institutions remain in the denominator or fall outside a record-level population. Verify is never converted to a definitive status.
A blocker wins.
Same-specificity records with conflicting approved meaning are withheld by the resolver, so neither enters the dataset. Narrower published scope can override a broader record only for a matching query. A source conflict or changed fingerprint triggers re-verification and a new public version when the result changes.
Topics select records; they do not change rules.
| Topic | Classification rule |
|---|---|
appliances | Assigned when the normalized subject and approved wording concern an appliance provision, prohibition, or use restriction. |
bed_size | Assigned when the normalized subject and approved wording include bed-size or bedding-size guidance. |
move_in | Assigned only to the normalized move-in-arrival subject. |
micromobility | Assigned to storage or charging rules for bicycles, scooters, or related batteries. |
preparedness | Assigned to emergency-preparedness packing guidance. |
prohibited_items | Assigned when the subject is a list of prohibited items or equipment. |
room_provision | Assigned when the subject describes an item or furnishing supplied in the room. |
What version 1.0.0 cannot establish.
- The release contains three California public universities selected from current Dorm, Sorted. coverage; it is not nationally representative and should not be used to estimate all colleges.
- Topic coverage is uneven because official housing pages publish different details and the dataset does not treat silence as policy.
- Most records are institution-level or population-level. A residence hall, room type, assignment email, accommodation, or later official notice may be narrower and controlling.
- The snapshot supports descriptive record counts, not causal conclusions, time trends, or policy-change frequency. One release is insufficient for a change-frequency analysis.
- No comparative chart is published for version 1.0.0 because three institutions and uneven topical coverage are inadequate for a representative visual comparison. Every supported analysis is provided as text and an underlying table.
One validated public source.
The public JSON is the validated source for the HTML report, methodology page, CSV, and Dataset structured data. The generator filters the campus-rules resolver at the fixed cutoff. Contract tests rebuild every analysis and the CSV and fail if any public figure, row, or metadata field disagrees.
Corrections create a traceable next step.
Report the dataset version, record ID, disputed wording, and an official university or housing source when available. A report never edits the dataset directly; it triggers the same fail-closed verification process and, if needed, a new immutable version.
Do not send names, student records, housing assignments, access codes, roommate details, or screenshots containing private information.
Read the public corrections process