Where NYC 311 quietly gives up.
Every 311 request ends with a one-line resolution, and a lot of them say some version of "we looked, we didn't see it, closed." This page counts those, for every city agency and community board, and tracks how often the same problem is reported again at the same address within 30 days.
Your block or community board
Addresses are matched against the top re-reported addresses list and highlighted below; community boards are matched by number and borough.
Who closes without looking
Where it happens
Community boards are the 59 local districts, plus a few catch-alls the city uses for unassigned requests. Requests are attributed to the board the city recorded.
What gets ignored
Only types with at least 200 requests in the window, by the agency that handled them.
The problems that keep coming back
The 20 addresses where the most closed requests were followed by a new report of the same problem within 30 days.
What the numbers mean
Closed without action
Share of closed requests whose resolution text says the agency did not observe the problem, found no violation, could not get in, could not find it, decided it was a duplicate, sent it to another agency, or otherwise closed it without describing a fix or enforcement. It is a text heuristic, not an official category: the city does not publish one. A high rate is not proof of neglect. Some complaints really are unfounded, some problems go away on their own, and noise complaints in particular are often over by the time anyone arrives. But a district or agency that sits far above the citywide rate is a place where reporting a problem is unusually unlikely to lead to anything.
Re-report rate
Of requests that were closed, the share that were followed by a new request at the same address for the same complaint type within 30 days of the close. This is the closest public signal for "it was not actually fixed." Only closes that had a full 30 days of data after them are counted, so the most recent month of closes is excluded from this rate.
Matching is on the address string exactly as the city recorded it, so the rate undercounts problems reported with slightly different addresses or as a different complaint type. It also counts repeat callers: a handful of addresses file the same noise or parking complaint hundreds of times, which pushes the citywide rate up, especially for NYPD. Compare agencies and boards against each other rather than reading the absolute number as a fix-failure rate.
Median hours to close
Half of closed requests were closed faster than this, half slower. A very short time (under an hour) usually means the request was closed at intake, for instance as a duplicate or a referral, rather than after someone went out to look.
Methodology and caveats
Data: NYC Open Data, 311 Service Requests (erm2-nwe9),
requests created in the window shown at the top of the page. Everything is recomputed from scratch by build.py; there is no database and no manual editing.
A request is closed when its status is "Closed" and it has a closed date. It is closed without action when the resolution description matches any of these phrases:
…unless it also contains a phrase indicating something was done, which overrides:
Known limits: the window is 90 days, so seasonal complaint types (heat, hydrants) skew the picture; agencies with a lot of intake-level closes (duplicates, referrals) are counted the same as agencies that send an inspector; "no violation" for a police noise call and "no violation" for a housing inspection are very different events but both count. Agencies and boards with fewer than 100 requests are hidden. Requests with no address are excluded from re-report matching.
Source code and the build script are in the zohran repository.