Methodology

This tool began as research infrastructure for a senior thesis studying neighborhood change in Chinatown and Two Bridges. NYC development information is spread across several independently structured public datasets -- DOB permits, PLUTO parcel data, ACRIS property records, Census ACS -- which makes repeated parcel-level research cumbersome. It combines those sources into one searchable interface and tracks new activity automatically, built for a single researcher rather than as a general-purpose real-estate platform.

It's deployed under the name Division, for Division Street -- the street used to hand-split the official Chinatown-Two Bridges neighborhood area into the two separate study areas this tool tracks (see “Study geography” below).

The study area covers Chinatown, Two Bridges, and the adjacent Lower East Side. These boundaries are researcher-defined, not official city administrative geography. They start from NYC's 2020 Neighborhood Tabulation Areas (NTA), with the combined official Chinatown-Two Bridges NTA hand-split along the Division Street centerline -- the conventional dividing line in local usage -- into separate Chinatown and Two Bridges areas; Lower East Side is taken as-is from its own NTA.

Which parcels fall inside those boundaries is itself a judgment. Most are resolved by testing whether a PLUTO parcel's centroid falls inside a study-area polygon. A small number of parcels have no PLUTO centroid at all; those are admitted instead when their tax block unambiguously belongs to one study area (a weaker claim, recorded per parcel). Permits are matched the same way in principle: most by their parcel's BBL, and a small remainder -- lots the BBL allowlist can't reach, such as condo unit lots or merged/demapped lots -- by their own point falling inside a study-area polygon instead. A permit matched this second way has no parcel to attach to and so has no parcel page, even though it appears in the feed and on the map.

• DOB NOW (NYC Dept. of Buildings) -- current permit filings and their approval/issuance status.
• PLUTO (NYC Dept. of City Planning) -- parcel-level land use, zoning, building characteristics. A pinned, versioned snapshot, not a live feed.
• ACRIS (NYC Dept. of Finance) -- recorded property transactions (deeds, mortgages). Not yet loaded into this tool.
• Census ACS -- neighborhood-level demographic and housing context. Not yet loaded into this tool, and will never be attached to an individual parcel -- only to a neighborhood as a whole.

Ingestion is intended to run once daily via a scheduled job. DOB NOW permits are fully reloaded on every run, since a permit's status can change after it first appears and re-scanning also catches newly-matchable spatial cases; PLUTO is a versioned snapshot reloaded as a whole when a new version is pinned. Every ingestion run is logged, and a run that fails is recorded as failed rather than silently skipped.

The Borough-Block-Lot number (BBL) is the primary key tying permits, parcels, and property records together. Each source keeps its own native record ID alongside the BBL for traceability back to the raw government record. Census data is joined at the census-tract level, never to an individual parcel.

New building

A filing for new construction on the lot.

Alteration

A filing to modify an existing building -- the large majority of activity.

Demolition

A filing for full demolition of a structure.

Other

Filings that don’t resolve to one of the three categories above (e.g. "No Work" filings).

• Study-area boundaries are a researcher-drawn approximation, not official geography.
• A small number of parcels and permits are matched to the study area by a weaker method (block membership or point-in-polygon) than the primary centroid/BBL match; this is recorded per row rather than hidden.
• ACRIS property records and Census demographic context are not yet loaded -- parcel pages currently show permit history only.
• DOB legacy (pre-2016) permit history is not yet loaded.
• Estimated project costs are self-reported by filers at time of filing and are not independently verified.