NYC Open Data Civic Toolkit

This tab crosses two public datasets to find properties under financial and physical distress at once: behind on property taxes, and sitting on unresolved building violations.

The two signals

• Behind on taxes: the DOF Tax Lien Sale List (9rz4-mjek) — the properties the city has noticed as eligible for its annual lien sale because of overdue charges. Only liens that include property tax are counted (water_debt_only is not water-only). That column carries TWO vocabularies across cycles, N/Y in the 2025-06 batch and NO/YES in earlier 2025 batches, so both are matched; filtering on one spelling silently drops whole cycles. Only the single most recent cycle is counted, and it is discovered at run time rather than assumed, so a newly published cycle can never be blended with an old one in which a lot has since paid off. The cycle in use is stated above the table.
• Open building violations: active DOB/BIS violations (3h2n-5cm9, category text containing "ACTIVE") plus active ECB violations (6bgk-3dad, ecb_violation_status = ACTIVE). ECB carries the unpaid penalty (balance_due), summed per property.
• Distressed = a tax-delinquent property that also has at least one open violation of either kind.

Joining them honestly

• The datasets are joined on Borough-Block-Lot, the city's parcel key. The raw block and lot strings are formatted differently across datasets (the violations feeds zero-pad, e.g. 00847/00038; the lien list does not, e.g. 8026/130), so a naive text match silently finds nothing. The keys are normalized (leading zeros stripped) before joining. Placeholder and TEST lots are dropped, and the join uses BBL rather than BIN, which is frequently a 0000000 placeholder in the DOB feed.

Caveats: the lien-sale list is a noticing-eligibility list, not a ledger of exact back-tax balances (NYC does not publish per-parcel tax arrears), and a property drops off once the debt is sold or cured, so this reflects the current cycle, not all-time delinquency. Open-violation counts are point-in-time. The ECB balance_due is the penalty owed on those violations, not the tax owed. Source datasets verified live 2026-09-16; producer scripts/aggregate_property_distress.py.

Resurfacing a street is two separate visits. A milling crew grinds the worn asphalt off, leaving a grooved surface with the manhole covers standing proud of the road. A paving crew returns later to lay the new course. The gap between the two is where the complaints live.

The source

• Street Resurfacing Schedule (xnfm-u3k5) — DOT publishes the blocks its crews are scheduled to mill or pave, refreshed weekly. Every row is one block, one day, one work type, with the crew (in-house or contract) and shift (day or night).
• Every view on this tab queries the dataset live, so it reflects whatever DOT published most recently rather than a copy taken when this page was built. There is no precomputed file and nothing that needs regenerating.

The lookahead is short

• DOT's schedule runs only a few days ahead — typically to the end of the current week. There is no months-out plan in this dataset, so "when is my street due" is answerable for the next few days and no further. If a street is not listed, that means it is not on this week's schedule, not that it is not planned.

Milled but not paved

• Each block carries DOT's own segment id (oftcode). A block is counted as open when its most recent milling is newer than its most recent paving (or it has no paving on record in the window): ground down, not yet resurfaced. Only completed work counts — DOT publishes a few days ahead, and a paving booked for Thursday does not make the street paved today. Any work already scheduled shows in the Coming up column instead.
• For blocks that did get repaved, the typical milling-to-paving turnaround is 16 days (median across 1,852 blocks repaved inside the window). That is the yardstick for whether an open block is simply in progress or genuinely lagging, and it is recomputed from the live data on every load rather than quoted from a constant.
• The backlog is bounded to the last 90 days. That is not an arbitrary cut: with a turnaround measured in weeks, a block still "open" after three months is far more likely to be a paving DOT never logged than a street genuinely left in pieces. A full-history rollup surfaced 5,611 such blocks, and presenting those as a real backlog would have been misleading.

Caveats: this is a work schedule, not a completion log. A row means crews were planned to be on that block that day, and weather or equipment routinely moves the work, so scheduled is not the same as done. That cuts both ways in the backlog: a block genuinely paved on a day that never got its own row will still look "open" here. Treat it as a list of candidates to check, not a proven inventory of unpaved streets, and note it cannot show a block open for longer than the 90-day window. Gaps are calendar days, measured against today in New York rather than the reader's timezone. Coordinates are published in NY State Plane feet and reprojected to WGS84 for the map. Every figure on this tab is queried live — the schedule from xnfm-u3k5, and, in the milled-not-paved view, completion from DOT's protected-streets feeds wyih-3nzf (blocks) and bryy-vqd9 (intersections), all verified 2026-07-21. Nothing is precomputed and there is no snapshot to go stale. The street-history view reads the schedule only, so it reports what was booked and does not check whether a paving was confirmed; the milled-not-paved view is the one that checks.