Five things it refuses

Five readings this record will not support.

1. That the city is generous

The reading. Four places in five set no limit. Seventy-one ask nothing at the door. Eighty-nine have a free chair. A stranger is welcome almost everywhere.

Why it fails. The record measures what is stated, not what is done. Nobody was moved along in two hundred and six visits because nobody stayed long enough to find out, and a hundred and sixty-four of those visits were in the quietest hours of a weekday.

Twelve of the eighty-nine free chairs exist so that somebody may sit; the other seventy-seven are for waiting or incidental to something else. And in a hundred and forty-one places nobody said a word to a stranger. To compare observations across locations, reviewers can explore the current reporting model and define the decision, access rules and review date in advance.

What survives. That very little is stated, and what is stated is mild.

2. That the city is mean

The reading. A hundred and seventeen places have no free chair. The commonest kind charges for the seat in ninety-five per cent of cases. People assume they are on a clock in half the city. A separate perspective on streets, seating and urban observation is available from RNIB.

Why it fails. The assumption is the observer's, not the city's. Of the places themselves, eighty-two per cent limit nobody, and the free third of the record is larger than most people guess.

What survives. That sitting is conditional on a transaction in more than half the rooms, and that the conditions are almost never stated.

3. That free seating produces company

The reading. Provide chairs and people will be among each other.

Why it fails. Twenty-five per cent against twenty-one, on eighty-nine places and a hundred and seventeen. The record looked and found nothing, across eight comparisons on three fields.

The three kinds where conversation does happen — pubs, places of worship, halls — are scattered across every value of every field, and share nothing the walk measured.

What survives. That free seating is an access finding and not a sociability one, and that the two are usually claimed together.

4. That the figures transfer

The reading. Twenty-three per cent of places produce conversation; eighteen per cent limit staying; half of all assumptions are wrong. Useful benchmarks.

Why it fails. One city, one season, one observer, forty minutes a room, selected by what could be walked to. Fifty-eight cafés in one city are not fifty-eight independent observations of a café: many share owners, rents and a population.

What survives. That the patterns are real in these rooms, and that the columns bear on each other as described.

5. That half the city imagines a rule

The reading. A hundred and five phantom limits. People walk around believing they are restricted when they are not.

Why it fails. Nobody was asked. The assumption field is one person's, recorded honestly under a fixed rule, in every place, including where being wrong was embarrassing — and that is all it is.

Whether anybody else makes the same guesses is untested, and it is the first question a second walk should ask.

What survives. That one walker, guessing in a fixed way before checking, expected a limit in a hundred and twenty-eight places and found one in twenty-three.

6. That any kind of place is better than another

The reading. Pubs produce conversation fifty-five per cent of the time and museums none. Something is being ranked.

Why it fails. What that sequence sorts is purposes, not merits.

A museum is for looking at objects, and people are quiet in it by design. A library is for reading. A concourse is for leaving. None of the three is failing at anything; they are being measured against a test they were never built to pass.

The record applied one narrow test — whether an exchange happened between strangers in forty minutes — and a room is not summarised by it. A person wanting to be alone among others is well served by a hundred and forty-one places here.

What survives. That the kinds differ enormously on this one measure and that nothing about their terms explains it.

What the six refusals have in common

Four fail on the instrument: nothing tested, nobody asked, forty minutes a room, one pair of eyes.

Two fail on the sample: one city, one season, and places that are not independent of each other.

None fails because a count is wrong. Every figure on this site is a count of what was recorded, and what they cannot do is carry the sentences a reader will want to build from them.

The arithmetic is sound and the reach is short. This page is the difference between the two.

What the five have in common

Three fail on the instrument: nothing tested, nobody asked, forty minutes a room.

Two fail on the sample: one city, one season, and places that are not independent of each other.

None fails because the figures are wrong. Every count on this site is a count of what was recorded. What they cannot do is carry the sentences people will want to build out of them, and this page exists so that the wanting is answered in advance.