Building Biography gives every tax lot in New York City a page: its permits, violations, complaints, sales, owners, inspections and 311 requests, stitched from about thirty city datasets into one timeline, with an A to F grade for residential buildings computed from a published formula. The Mice Map plots every actively inspected restaurant in the city at its front door, carded with its health grade and whether an inspector recorded evidence of mice.
Every record on both sites comes from NYC Open Data, the city's own published records: the Department of Buildings, Housing Preservation and Development, the Department of Finance, the city register (ACRIS), the Department of Health, the Fire Department, the Department of Environmental Protection, 311, the city's tax-lot map (PLUTO), and the Department of City Planning's Neighborhood Tabulation Areas (the neighborhood boundaries). Nothing in the record is typed in by hand and nothing is scraped from private sources. Every number regenerates from the source each week; every page says which data date it was built from. The one exception is the reader poll on building pages ("Lived here? Would you rent here again?") — an unverified yes or no from site visitors, shown separately from the record and never used in a grade.
One person builds it. Raw records are ingested as published, normalized and quality-checked table by table, joined on the city's building identifiers, annotated, and served. The joins are the hard part. A single building appears under a tax lot number in one dataset, a building identification number in another, a street address spelled six ways in a third, and a condominium's billing lot in a fourth; resolving those into one record is most of the work. The grading method is published in full, and the landlord rankings use the same formula weighted by units.
A page here is a record of what the city has published, not a verdict on a building or a business today. Records lag reality by days or weeks, agencies correct their own data, and a violation that has since been fixed still appears in the history. When a data source is unavailable the page says nothing rather than implying a clean record. Grades measure recorded distress relative to similar buildings; a small building with one bad year can grade badly. Restaurant closures and grades are shown exactly as the Health Department published them, including the date, so you can judge how current they are. The reader poll on building pages is the one piece of a page that isn't a public record — a self-reported opinion, kept visibly apart from the record, and it never affects a grade.
The reader poll on building pages stores a random identifier your browser keeps, hashed together with a server-side key separately for each building, so one reader's answers can't be linked across buildings; a hashed network prefix for the day, used only to limit how many new answers one connection can add in a day; and your answer itself with the day you gave it. It does not store your IP address, your name, or any free text.
Your current answer stays on file until you change it; the daily rate-limit hashes are cleared once the day has passed. The poll uses Cloudflare Turnstile to block automated submissions — it loads only once you touch the poll, and its own handling of what it collects is described in Cloudflare's Privacy Policy.
If a page shows something that is wrong in the city's records, the fix has to happen at the source: the agency that published it. If a page misreads a correct record, that is a bug here, and an email to support@mice.nyc with the page link will get it fixed. Names of registered owners and agents are shown as filed with HPD.
This is a one-person side project, free to use, with no advertising and no accounts. Servers, data storage and the weekly refresh cost money; if it has been useful, you can chip in whatever you like.
Built on NYC Open Data, DuckDB, Cloudflare Pages and R2, MapLibre, and the Public Sans and Source Serif typefaces. New features are listed on the What's new page.