Ask anything about a city

A model states last year's opening time with this year's confidence, and you would act on it. Hedging everything is the same as hedging nothing, so every claim comes back sorted by how fast it decays — a property of the claim, not of the city.

• durable — fixed by physical geography, history or built form: the river, the ridge that forced everything into one valley, the grid imposed after the fire, why a district feels the way it does. It does not become false because a year passed.
• slow-drift — true over years and drifting over years: population, cost, the broad shape of the transit network, the direction a district is moving. Every one carries an as-of year. A drifting figure without a date is the thing this app exists to stop.
• current-state — the present operating status of a specific named thing: prices, fares, admission, opening hours, timetables, service frequency, whether a business or line is still running, construction, entry and visa conditions, exchange rates, event dates. These are never stated — not with a hedge, not approximately, not "as of my knowledge".

In place of a volatile fact you get three things: the durable structure around it (how timed entry works at all, why the airport line has always been the slow one), the shape of the answer without the number, and the named authority that actually holds it — the operator's booking page, the transit agency, your government's own entry-requirements page. "Look it up online" is not an answer; naming who holds it is.

A claim that arrives with no stability label is not quietly treated as durable. It is reported to you as an unclassified claim, which is the defect it is.

The city is named in your input, so one retrieval grounds it before the run: the page fetches Wikipedia abstracts and grounds the city's anchor facts — country, region, population — against them.

That is not enough on its own. A search result is a snippet, not an infobox, so a "cite only retrieved records" contract can still be satisfied by citing a real page for a fact the page never mentions — a confabulation laundered behind a real URL, with the verification passing. So the page re-reads every cited abstract and checks it literally contains the value claimed, and reports back any citations the source does not actually support. A citation that fails is shown to you as a false citation, more prominently than an honest "model knowledge" would have been, because false evidence is worse than absent evidence. Each snippet is printed in full underneath so you can adjudicate the finding yourself.

Districts are described in their own terms. This app ships no closed vocabulary of neighbourhood types, deliberately: one built on European and North American cities describes Lagos, Chongqing, La Paz or Ahmedabad confidently and wrongly — where the wealthy live on the periphery, where altitude sorts people by class, where the informal settlement is the economic engine, where there is no single centre at all. The page also flags descriptions that map a district onto a district of another city — "the Brooklyn of X", "the Venice of the North" — because those tell you about the reference city and import a social structure that may not apply.

It will not describe a city it cannot identify, and it will not silently merge two cities that share a name.

Runs are metered. The line above the fold of this panel shows what the run you have configured would reserve, in US dollars, before you commit to it — a reserve is a hold, not a charge, and the unused part of it is released when the run settles.

Depth is the main lever: Quick reserves least, Deep most. The worked example costs nothing at all — it is a saved run bundled with the page, so it needs no account and spends nothing.