It was two days before I closed the East Asia base camp.

I had a little time, so I went to see The Odyssey. Coming out of the cinema, I walked about twenty minutes back to where I was staying. The street was dark, the lamps were sparse, and it was quiet.

I thought the walk was lovely. I don't walk like that in Seoul. I don't walk like that in New York. I walked like that because it was Jeju.

Two mornings later, in a project meeting, the people on the other side of the screen told me the news. Missing-person reports had been closed without anyone checking. Searches had restarted. People were being found.

From that moment Jeju was a dangerous place to me.

Nothing in Jeju changed between the night I walked and that morning. Nothing happened to me on that street. What changed was not the street. It was what I knew.

And I will not walk that street the same way again. Showing me the statistics will not fix it.

That is what this letter is about. Not what happened in Jeju, but the harder question underneath it: what is a city's safety actually made of?

Jeju is not, statistically, a dangerous place. More than 99% of missing-person reports are resolved, and the number of reports filed has been falling.

The failure was somewhere else. A single officer could close a case, and no layer existed to verify that judgment. Reports closed without any check were counted, in the statistics, as resolved.

So the 99% is not a lie. What was never established is what it proves.

This is where the story stops being about Jeju. Cities that publish safety figures drawn from unverified systems are not rare. Jeju is simply the one where somebody looked.

And travellers were never reading the figures anyway.

Travel decisions are not risk calculations. They are uncertainty avoidance. Nobody computes the probability of something happening to them here. What they do, instantly and without arithmetic, is ask: if something happened to me, would this city come looking?

The first question can be answered with data. The second cannot. That asymmetry is why a safety crisis always outlasts the danger that caused it.

So how have other cities handled this?

London · Shanghai · Mumbai · Paris · Bogotá · Mexico City

Britain's answer fits in one sentence. Closing a case is not one person's decision.

A report that does not become a formal missing-person record must be closed with a bespoke closing code, and the stated purpose is written into the guidance: so that reports which never became formal records can be monitored and audited afterwards. Everything filed under "nothing to see here" is collected and reviewable.

Some force policies go further. Where the person has not been located, the incident log cannot be closed at all without first creating a missing-person record. Authority to discontinue sits with senior management, not the officer holding the case. Risk assessments are endorsed by a duty inspector, and every change of risk level requires a fresh assessment with the rationale recorded.

There are clocks, too. If a person has not returned within 72 hours, case details go to the missing persons unit within the next 12. Unidentified bodies must be reported to the national unit within 48 hours of discovery. And it is audited annually, against a sixteen-question standard.

The open question — audits are annual and sampled. Do they catch deliberate falsification in time?

Shanghai went the opposite way. It treated safety as a problem of seeing.

One district shows the scale. Xuhui tripled its facial recognition network, adding 2,500 cameras to an existing 1,200 — roughly 175 cameras per square mile across 21 square miles. Officials estimated the system would capture and analyse 25.9 million faces per day. That is 18,860 people a minute.

The cameras are the smaller half. The expansion was built to mine files covering more than 50 million residents, assembling profiles of movement, social relationships and activity, with AI flagging behavioural deviations to trigger early warnings. Cameras moved past public space into residential compounds; in some buildings they were installed in lifts and residents only found out afterwards.

China operates an estimated 170 million-plus surveillance cameras. In one well-known case, police picked a wanted man out of a crowd of 60,000 at a concert.

And this system does not find missing people.

A Le Monde report in April 2026 is the crux of this entry. The infrastructure now responds almost instantly to minor infractions — crossing a line, running a light on a scooter — issuing near-immediate fines. On missing persons, it does not deliver.

That is not a limit of the technology. Surveillance reflects what a city has decided to look for. It does not tell a city what it should be looking for.

The open question — why does the most capable city fail at this? And is the price worth paying?

India built half of the thing this problem needs, and the half it skipped is the interesting one.

What exists is real. The national Digital Police portal lets any citizen run a Missing Person Search directly, against a consolidated database that holds missing persons and unidentified bodies side by side. Maharashtra runs its own citizen portal with the same search. District police dashboards list Track Complaint Status, Missing Person and Unidentified Dead Body as separate public entries.

What does not exist is the other direction. To learn what has actually happened to a case they filed, a complainant must lodge a fresh complaint at the same station — or file a Right to Information request.

So citizens can see who is missing. They cannot see what is being done about the person they reported. For that, you petition the state.

The open question — searchability got built. Why didn't traceability?

Paris is the only city here with a timetable.

The fall was steep. Visitor arrivals dropped by roughly half in the immediate aftermath of November 2015. Across 2016 the region lost more than €1 billion in tourism revenue, about 6%. The Louvre, the Musée d'Orsay and Disneyland Paris each saw visits fall between 9% and 13%. Three million fewer visitors over the year. Twelve consecutive months of decline — the sharpest since 2009.

The recovery was real too. January 2017 brought over 1.5 million hotel guests, a ten-year high for the month. The first half of 2017 recorded 16.4 million arrivals in Île-de-France, the strongest opening half in a decade, with foreign visitors up 14.9%.

But the useful finding is not the speed. It is the asymmetry.

Domestic spending barely moved and recovered fully within twelve months. Foreign visitors took far longer, and not evenly. By early 2017 Belgian, Spanish and Swiss arrivals had returned, while British numbers were still down 3.4%, Chinese down 6%, and Japanese down 19%. Recovery ran faster in France's provincial cities than in Paris itself.

The closer you are, the sooner you come back. The further away, the longer it takes. That is distance in kilometres and distance in information — because to someone far away, the city exists only as news.

The open question — recovery happened. Nobody has separated how much was the response and how much was simply time.

Bogotá has spent two decades fighting a different war: the one where the indicators improve and perception refuses to follow.

Reported theft fell from 159,000 to 130,000, down 18.2%. Residential burglary fell from 8,200 to 6,000, down 26.6%. And 58.3% of residents still say they do not feel safe. In other surveys, more than 80% say they live in an unsafe city.

Public transport is the most honest case. More police were deployed on the TransMilenio system and actual safety improved. 60% of riders still perceive it as unsafe.

Part of the cause turned up in the survey itself: roughly half the crime information residents receive comes from television. Cases were documented where incidents that occurred in Brazil months earlier were reported as though they had happened in Bogotá.

What Bogotá did was refuse to leave perception unmeasured. The Chamber of Commerce, working with the city and the national police, surveys around 8,200 residents across 19 localities every six months, with a 3% margin of error. It is one of very few cities that manages safety as a perception indicator and not only a crime statistic.

The open question — measuring the gap worked. Closing it has not.

This is not a solution. It is a destination.

More than 135,000 people registered as missing. More than 230 victims' collectives. And the most trusted searchers are no longer state institutions — they are the families.

The mothers developed their own field method: disturbed soil, insects that should not be there, a patch of unusually green vegetation. Knowledge nobody had ever written down.

Recently the direction reversed once. Jalisco's State Search Commission began a project to give that knowledge scientific backing. The commission's head put it plainly: the work started from what the searching mothers knew, and what the state added was technical support for something already proven in the field.

The cost is severe. Official figures record eight searchers murdered across 2022 and 2023. An Amnesty International survey found 97% had experienced violence or harm in the course of searching.

The open question — has any city reversed the path once citizens took over?

Put the six side by side and a shape appears.

London fixed the record. Shanghai pushed observation to its maximum. Mumbai opened access halfway. Paris measured time. Bogotá began measuring perception. Mexico City shows what remains when all of it fails.

Nobody solved the whole thing.

London's audit is annual and sampled. Shanghai's cameras are pointed at something else. Mumbai's portal answers who is missing but not what happened to your report. Paris cannot separate its response from the passage of time. Bogotá measures the gap without closing it.

And no city has combined them.

What's needed is not another policy. It is somewhere to test one together — a city with countable points of entry and exit, where an experiment can be measured, at a scale where failure is survivable.

Last weekend's edition set out those conditions. Jeju has all three.

So here is the proposal, and it is where the next edition begins.

Practitioners, developers and researchers from six cities, each bringing the problem they could not solve at home, designing together in Jeju.

The output would not be a report. It would be working prototypes. Four candidates are already obvious.

Report status, visible to the person who filed it. An interface where a citizen can see what actually happened to their own report. An institution proving itself and a citizen verifying an institution are two different kinds of safety.

Closure verification. Mandatory evidence attached at close, and an alert when what was entered doesn't match what was filed. London's audit moved from sampled to real-time.

Perception instrumentation. Bogotá's six-month survey, run more often and more cheaply. No visitor-facing version exists anywhere.

Early-detection protocol. Location-tag based. What Shanghai's cameras could not do, approached from another direction.

If Jeju hosts this, Jeju stops being the city where something became known. It becomes the city where the world works on it.

That is a kind of trust no campaign can buy.

Format, timing and participation are still being worked out. The next edition takes this up in full.

bcdW Current — one city's today, through the eyes of other cities.

One City. One Story. Many Views.