About
What this is
This online platform forms part of the research project Managed Time: Reading the Public Record of Mott Haven by Hana Soule. The project was developed at the Centre for Research Architecture (CRA), Goldsmiths, University of London, as part of the MA Research Architecture program (2025–2026), with the guidance of the studio tutors. It accompanies a dissertation and an installation shown as part of Research Agency, a group exhibition, 24 September–2 October 2026. Each figure gives its own source and extraction date in the margin beside it.
This is an investigation into the instruments through which Mott Haven is governed. Mott Haven is a neighborhood at the southern tip of the Bronx, across the Harlem River from Manhattan, in the district the city numbers Bronx Community District 1. The instruments are the permits, orders, map changes, designations, special permits, awards and filings by which agencies act on land and on the people living on it. The City files its records by district, block and lot. The conditions this project follows, its noise, its dust, the time people spend waiting, do not stay inside those units, and the gaps in the record are where they cross them. The scales remain available as an index. At each one it asks the same question. What did the instrument authorize, who issued it, how long did it last, and what did it leave in the record?
The occasion for the work is the jail under construction in Mott Haven, one of four facilities intended to replace the jails on Rikers Island. But the subject is not the jail. It is the administrative apparatus that sited it, and the fact that the same apparatus — the same agencies, the same map, frequently the same instrument — also governs housing on the blocks around it.
Position
I provide research and policy support to No Brooklyn Jail, a campaign with roots in the No New Jails network. This project began in that work. The City is replacing one jail system with four, and through the same instruments it meets needs for housing, mental health care and social services with carceral space: secure treatment beds counted alongside jail beds, shelters whose hours each facility sets and where entry and continued residence are both conditional,1 a jail designed to read from the street as a civic building. The violence the federal court has supervised on Rikers since 2015 belongs to that reliance on detention, and new buildings carry it with them.2
How the work was done
The method is the same at every scale. Find the instrument; establish what it authorized, who issued it, how long it lasted and what it left in the record; then set what an institution committed to in public against what its own records show was done.
In practice that means reading documents that were not written to be read this way. Land use applications and the commission reports that decided them. Community board recommendation forms, with their tallies. Environmental review documents, and the technical memoranda that amended them after the fact. Exhibits filed in litigation. Registers of ownership, of assessment, of violations and of enforcement. Building footprints, including the ones the city retires when a building comes down. And, on the jail site itself, the environmental monitoring record: the authorizations that set when work may happen and for how long, the plans that govern dust while the ground is open, the vibration monitoring an excavation owes the buildings beside it, and the monthly reports that record it.
That monitoring record was the way into this project, and it set the question the rest of this project asks. Read one at a time, the documents are unremarkable — a two-week permission to start an hour early, filed and granted and expired. Read as a series, they show those two weeks renewed again and again without a gap. They also show which of the things a neighbor notices the apparatus can turn into a permission at all, and which it cannot: noise has a term and a renewal, vibration has a plan, dust has neither.
What a monitoring record registers, and what its reports publish, is not the same as what happens on a street. That distance is a property of the records, and it is what this project is built to show.
“Embedded presence. Active presence in communities. Uplifting community voices and centering community expertise. And really mitigating any kind of extraction — offering to provide more than receive. This work doesn’t have to be so transactional. It can be humane. It could be intentional. It could be genuine.”
How the figures are made
The records cited here were extracted between 12 August and 25 September 2026. Where an extract carries no retrieval record, the file’s own date stands in for it.
Every number published here has an entry in a figures register recording its source, the basis on which it was derived, and the date it was established. Where two figures should not be read together, the register records that as well, and they do not appear on the same page.
Numbers are reported for the geography that produced them, and that geography is named. A figure computed for the study area is labeled as the study area; a figure published for the community district is labeled as the community district; survey estimates are reported for the geography that publishes them, whether that is a PUMA, a CDTA or a neighborhood tabulation area. The name of the place is used in prose for the place. It is never used as the label on a number.
Spatial analysis is carried out in the State Plane projection the source records use, and delivered in the projection the web requires; no figure is ever computed from a delivered file. Distances follow the source documents, in feet and miles. Nothing is measured in blocks.
Where a figure is uncertain, contested between sources, or resting on an assumption, it is shown with that condition attached rather than published clean and qualified elsewhere. Where two City documents give different numbers for the same thing, both appear and the discrepancy is reported as a discrepancy. The same holds where a modeled value and a measured one disagree: both are shown, and where a conclusion depends on which of them is used, that is said at the point of use. A disagreement is not resolved by preferring the more recent, the more official, or the more useful.
What administrative records register
Administrative records register administrative activity. They do not register what happened in a building, and the difference matters at every point in this work. A repair recorded is not a repair made. An enforcement action is not a condition. A permit authorizes work in certain hours; it does not establish that work occurred in them. An award is not a charge. A sign-off is not a completion date. A retired footprint is not a demolition date. A modeled level is not a measured one.
Two limits are named because they affect findings this project relies on. Where a preservation program produced no subsequent record, the record cannot distinguish work that was not done from work that was done and not recorded, and this project says so at the point of use rather than in a footnote. And figures are added or set against one another only where they answer the same question about the same universe over the same period: penalty figures held by different tribunals under different identifiers are never summed, and counts of exceedance produced against different thresholds, or over different numbers of monitored hours in a day, are never compared.
Some of what this project publishes is modeled rather than recorded. Where it is, the method is named at the point of use, along with the source of every term it takes and the tolerance it carries. A model of this kind estimates conditions on open ground. It cannot establish what any particular building experienced, and it cannot establish that anything which happened to a building was caused by the work. Nothing modeled is presented as a measurement, and nothing measured is adjusted to agree with a model.
Complaint data is treated as a record of complaining, not of conditions. It is never presented as a causal estimate, and it is reported separately from violations and from enforcement, never merged with either.
Testimony
Residents, organizers and professionals who spoke to this project appear in a register kept deliberately separate from the administrative evidence, set differently on the page and never blended with it. Testimony describes; the record documents; neither is offered as proof of the other. Where someone states a fact the project also states as a finding, that finding is sourced to the record independently.
Ambient recordings made in the field are checked for bystander voices before anything is published. Interview text is edited only to remove fillers, false starts and repetitions; the wording is otherwise as spoken.
Consent here is two permissions, not one. Consent to speak is not consent to be named, and consent to be named is not consent for the material to be used. They were asked separately, are recorded separately, and where only one was given only that one is exercised. Where consent covers speaking but not naming, the speaker is attributed by role or as a resident of the area, with no affiliation and no address, because affiliation and residence together can identify a person in a small community as surely as a name.
Who is named
Corporate and institutional entities are named. Public officials are named for acts performed in office. Individual landlords and property owners are named where their conduct is part of the public record, because ownership is a public fact and the asymmetry of naming only companies would misdescribe how property here is actually held. Organizers, academics and researchers are named where they have consented.
Tenants, occupants and family members are not named, except where they are parties to a public court proceeding and have been named in the press. No personal telephone number is published here. Numbers belonging to agencies, courts and businesses may appear in the documents reproduced here.
Certain addresses are withheld, and the withholding is systematic rather than incidental. Before any combination of details about a person is published, it is tested against the question of whether a neighbor, a landlord, a broker or an agency with a file could resolve it to one individual — not whether a stranger could. Where the answer is that they could, an attribute is withdrawn or the combination is not published. Those tests are recorded with their reasoning, so that they do not have to be argued again from scratch.
Two kinds of gap: silence in the record, and withholding
Gaps here are of two kinds and it matters which is which. Some are places where the record is silent — where no instrument exists, where an agency publishes an aggregate and not the underlying file, or where the thing that happened was never the kind of thing that generates a document. Those absences are findings and are marked as such, because what an administrative system does not record is part of what it does.
Others are deliberate withholdings by this project, made under the rules above. Those are marked too. No absence here should be read as an oversight without checking which of the two it is.
Some of what this project holds about the portfolio is not published. Certain details about its households, set beside one another, could identify individual people to someone with local knowledge. They were tested against that risk and do not appear together. The withholding is this project’s decision.
Sources and corrections
Primary records are cited to the document: the agency, the reference number, the date on its face. Company marketing and promotional material is not cited, including for background. Where a source is a pleading, it is treated as a party’s contention and used as a finding aid pointing at primary records, never as the evidence itself. Journalistic and advocacy sources are reported as what they said, attributed to their outlet and date. They are used to locate records and are never cited as evidence of a finding.
This project has made errors and corrected them, and the corrections are part of the method rather than an embarrassment to be tidied away. Where a figure, a date or an attribution published here has been changed, the change is logged with what it replaced, and where a line of argument was abandoned because the evidence did not support it, that is logged too. A record of corrections is the only honest evidence that anything was checked.