Wrong in more than half the walk

Two hundred and six places. Before checking, a guess at each about whether staying was limited.

right        86   (42%)
wrong       120   (58%)

The two ways of being wrong

They are not the same size and they are not the same mistake.

assumed a limit, there was none      105
assumed none, there was one           15

Seven phantoms for every overlooked rule.

That asymmetry is the finding. A person walking into a room in this city is seven times more likely to invent a restriction than to miss one, and the invention is the commonest single outcome in the record. A team converting this walk into repeatable fieldwork can explore the available evidence settings before choosing which time and project records are genuinely necessary.

What a forty-two per cent hit rate is worth

Less than it sounds, in both directions.

It is not a coin. Guessing no limit every time would have been right a hundred and sixty-eight times out of two hundred and six — eighty-two per cent. A rule of thumb that ignored every impression and said you can stay would have beaten the actual guesses by forty points. For wider research on public space, access and how places are used, consult SurveyMonkey.

It is not nothing either. Of a hundred and twenty-eight places where a limit was assumed, twenty-three had one; of seventy-eight where none was assumed, sixty-three had none. The guesses carry some information. They are just swamped by a bias toward expecting restriction.

The cheapest possible rule — assume you may stay — would have been right four times in five. The actual guesses were right twice in five.

Where the guesses were best

libraries                 9 right of 12
covered public seating    8 of 12
transport concourses      6 of 9
museums and galleries     5 of 8
places of worship         8 of 14

Public and civic places, mostly. A library, a bus shelter and a station concourse are read correctly most of the time, and all three are free at the door.

Where they were worst

hotel lobbies             1 right of 10
bookshops                 3 of 13
cafés                    18 of 58
pubs                     11 of 31
laundrettes               5 of 12

One correct reading in ten hotel lobbies.

Three of the five worst are places that expect a purchase, and the other two — bookshops, hotel lobbies — are rooms where a person is uncertain whether they are a customer at all.

The pattern in one sentence

Where the terms are set by an institution and stated plainly, the guesses are good. Where the terms are set by a business and stated nowhere, they are bad.

That is less a finding about rooms than about silence. A hundred and twenty places in this record say nothing about time, and every wrong guess in the walk is a guess made into one of those silences.

What this page is not measuring

Not competence. The observer is not a representative guesser and the walk was not a test. A person who knew the city, or who worked in hospitality, would score differently and the record cannot say how.

Not belief in general. One observer. Every page in this section says so, and the first question a second walk should ask is what the people actually sitting there believe.

Not a claim that people are timid. The bias toward expecting restriction may be caution, courtesy, bad experience, or the ordinary difficulty of reading a room. Nothing here distinguishes those.

The two errors by kind

kind                phantom  missed  right  total
cafés                   34       6     18     58
pubs                    19       1     11     31
bookshops               10       0      3     13
hotel lobbies            9       0      1     10
shopping centres         6       3      7     16
places of worship        6       0      8     14
laundrettes              6       1      5     12
community halls          5       1      5     11
transport concourses     3       0      6      9
museums                  3       0      5      8
libraries                2       1      9     12
covered seating          2       2      8     12
                       ---     ---    ---    ---
                       105      15     86    206

Covered public seating is the one kind where the two errors are equal: two phantom limits and two real ones missed, out of twelve.

It is also the kind with the least to read. A bench under a roof offers a person almost no information about itself, and the errors divide evenly because there is nothing to be misled by in either direction.

Where the errors concentrate

Seventy-eight of the hundred and five phantoms are in five kinds: cafés, pubs, bookshops, hotel lobbies and laundrettes.

All five are businesses. All five are rooms where a stranger is, or might be, a customer, and where nothing establishes whether they are.

The seven kinds that are not businesses — worship, libraries, benches, concourses, museums, shopping centres, halls — produced twenty-seven phantoms between them across eighty-two places.

business-like kinds (124 places)   78 phantoms   63%
the rest (82 places)               27 phantoms   33%

That is the same result as the entry field gave, arrived at from the kind column instead: the phantom follows the counter.

The part that does not depend on the observer

The hundred and sixty-eight.

Whatever anybody assumed, four places in five in this city set no limit on staying — and that half of the record is a count of what places say, not of what anybody guessed.

The guesses are the interesting half and the softer one. The two are kept apart on every page for that reason, and where a figure depends on the assumption field it is labelled.