Use our data

Everything on this site is generated from a few JSON files, and those files are yours. Journalists, advocates, researchers: download, analyze, republish, with attribution. Licensed CC BY 4.0.

The datasets

• airlines.json: 16 airlines. Economy seat widths by aircraft, seatbelt lengths, extender rules, customer-of-size and second-seat policies. The accountability view lives on the comparison page.
• parks.json: 5 theme parks, 75 rides. restraint types and manufacturers, test seats, modified rows, size language. Accountability on the parks page.
• transit.json: Amtrak, Greyhound, and FlixBus. Seat dimensions, seatbelt facts, second-seat policies.
• cruise.json: 4 cruise lines (Royal Caribbean, Carnival, Norwegian, Princess). Onboard attraction weight and size limits, tender-boat guidance, and shore-excursion caveats. Accountability on the cruise page. Disney Cruise Line is held pending our direct verification and is excluded from this file.

How it's verified: the badge ladder

Every data point carries its own source object: URL, access date, source type, and a confidence tier. From strongest to weakest:

• Verified by us: a named human read the source directly; the record carries their initials and the date.
• Official, pending our direct verification: recorded from the operator's own page via excerpt; queued for a human verification sweep (verification: "pending" in the JSON).
• Archived copy: the operator's own page, read from a dated archive because the live site blocks automated reading.
• Secondary: a named non-official source, with its publication date where it shows one.
• Direct correspondence: "told us by [operator] customer relations, [date]"; informal answers, never overriding published policy.
• Community reports: rider/traveler reports, dated, corroboration counted; used only where nothing better exists, and visibly badged.

Two standing editorial rules ride along in the files: the smaller/tighter figure always leads (the "planning number"), and conflicts are recorded both ways rather than silently resolved. Honest absences are explicit: not_published is a value, not a missing field.

How to cite

Any reasonable form works. Suggested:

SeatRuler airlines dataset, seatruler.com/data/, retrieved [date]. Licensed CC BY 4.0.

Each file also carries license, canonical_url, and cite_as fields, so a downloaded copy explains itself.

Freshness

Each file's generated field and per-source accessed dates tell you exactly how old every figure is. Re-download from the canonical URLs above rather than mirroring stale copies; the data updates on a published re-verification cadence (monthly for volatile items, quarterly otherwise).