The minimum number of movements.
Most analytical work in institutions stops at the first movement. A diagnosis is rendered, a slide is produced, and the work returns to the desk that requested it. The institution learns something true about itself and changes nothing, because no one is paid to close the loop the diagnosis named.
Our position is that observation without diagnosis is gossip, and diagnosis without closure is theatre. The three movements are the smallest set that obliges all three obligations at once: see honestly, name precisely, finish what you opened.
An observation is judged by its closure.
The bureau is not graded on the elegance of its diagnosis or the size of the dataset it processed. It is graded on a single quantity: the proportion of named loops that closed within the horizon agreed at the start.
If the law is not satisfied, the next observation begins by saying so. The bureau publishes its own miss before it publishes anything else. This is not modesty; it is the only honest way to keep instruments calibrated.
Six faculties, one instrument.
The platform that produces an observation is built around six faculties — not features, not tabs. Each one corresponds to a question an honest observer is obliged to ask before declaring anything.
- EyeSight Atlas What is in the city — every story, where it opened, where it is now, when it was last touched.
- EyeSight Anomaly What is different from itself — the points where the institution's record breaks its own pattern.
- EyeSight Numic What numbers we are willing to stand behind — the operations that survived audit, with the ones that did not crossed out.
- EyeSight Perspectives Who sees what — the same observation rendered for the desk, the boardroom, the regulator, and the citizen.
- EyeSight Spinfold The whole observation as a single object — rotated, narrated, and inspectable in one frame.
- EyeSight Closure What was moved and whether it closed — the written record of the institution's response, kept by the bureau.
Quarterly, then permanent.
An observation is not a report. It is a habit. The bureau's preferred cadence is quarterly, because the city moves slower than a sprint and faster than a year. A first observation establishes the Atlas; a second names the loops; a third closes them; and a fourth says, plainly, what closed and what did not.
By the fifth observation the bureau is no longer producing diagnoses. It is producing a record. The institution begins to speak of itself in the language the bureau introduced — Atlas, named loop, closure rate — because the language fits the work better than the words it had before.
A small institution, by design.
The Data Company of London is one chapter of an observational bureau. The chapter is small on purpose: a principal observer who signs every dispatch, a small bench of instruments-makers, and a longer ledger of correspondents. We do not scale the bureau by hiring; we scale it by opening another chapter, in another city, when the city has earned one.
We do not pitch. We do not run sprints. We accept commissions, in three forms — a single observation, a series of four, or a permanent seat — and we are slow because the work is slow.
Send a letter.
Most engagements begin with a single page. Tell us the institution, the loop you suspect is unclosed, and the horizon over which closing it would matter. The principal observer reads letters on Tuesdays. The first reply is usually a question, not a proposal: institutions are not yet legible enough at first writing to be diagnosed.
If the question is answered well, a first observation is scheduled. If the institution prefers, a worked example — the Spinfold — is available now, in writing, and the platform that produced it is open to inspection.
Read further, or commission.
Folio 001 — the founding note — sets out the why. The Spinfold shows the how on a worked example. Commission begins the engagement.