I wrote Sim Eternal City — Future City Storytelling Framework in Colombia. It was published this past May, and in the same month, after New York Design Week, I came to Jeju.

The first thing I noticed in Jeju was not a city. It was a slightly larger unit — the city-state. A political form most people assume vanished with antiquity, reappearing in near-future shape.

It was, perhaps, an unintended by-product. I had not come to Jeju looking for a city-state.

Since then I have listened to the problems raised on this island, and watched the solutions proposed for them.

People are leaving. The young do not stay. Empty houses accumulate. Tourists are plentiful; the money that stays is not. There is surplus electricity and a narrow path off the island. Waste has nowhere to go, the haenyeo are disappearing, and medical care requires a flight.

The diagnoses were accurate. The solutions were diligent. The people executing them were sincere.

And yet what remains is this: every solution went wrong at the same point.

The moment Jeju is placed as a part of South Korea, every Jeju problem is automatically translated.

Population becomes regional decline. Industry becomes local economic revitalization. Energy becomes grid interconnection. Tourism becomes tourism promotion.

Each translated problem arrives with a fixed menu of remedies: subsidies, corporate attraction, festivals, special zone designation, publicity.

These are not bad instruments. But every one of them is a tool for helping a lagging periphery catch up with a center.

There is a trap here. A solution premised on catching up locates its goal outside itself. If Jeju must catch up with Seoul, then Seoul defines success. The judge of whether Jeju is doing well stops being Jeju.

Read Jeju's problem list again and something strange surfaces.

Super-aging. Youth outflow. Surplus renewable energy with nowhere to go. Empty houses. Tourism overload. Freshwater dependency. Waste sealed inside a boundary. Knowledge disappearing before it is documented.

This is not the problem list of a backward province. It is the list that city-states will meet, in sequence, through the 2040s. Jeju is living all of it now, simultaneously, inside a clearly drawn boundary.

The numbers make it sharper.

Jeju's registered population peaked at 679,016 in August 2022, then declined for 35 consecutive months. April 2026 brought a 52-person uptick; by June it was falling again, to 662,792. Roughly 16,000 people gone in four years.

What matters more is the composition. In 2025, 53.8% of net out-migration was people in their twenties — 2,298 of them. Over the same period the 65-and-over share reached 20% and Jeju crossed into super-aged status.

When half of those leaving are in their twenties and the remaining structure ages, the problem is not that the population is shrinking. It is that the population's time structure is inverting.

Singapore has not yet faced renewable surplus. Dubai has not yet met super-aging. Amsterdam fights tourism overload, but people leak past the city boundary, so what worked cannot be isolated.

Jeju meets all of it inside a closed boundary, at once.

Epidemiology has a name for a sample like this. The early-onset cohort. Not behind — first.

And an early-onset cohort is an object of study before it is an object of treatment. This is the hinge. Read Jeju as a province and it becomes a place to be supported. Read it as an early-onset cohort and it becomes a place the world needs to learn from.

If what Jeju lacked were budget, this essay would be unnecessary. What it lacks is a narrative that reads Jeju as a complete unit.

Sim Eternal City is a tool for that narrative. Not urban planning but a future-city storytelling framework — a grammar for describing a city as an independent unit with its own timetable and its own definition of citizenship, rather than a subdivision of a nation.

It has four axes, and each becomes a test item in Jeju.

Empty houses and gotjawal. What changes when a city's first instrument is re-occupation rather than new construction? Every shrinking city on earth is waiting for this answer.

Can it plan on generational cycles rather than budget cycles? Jeju has a long-running discipline of self-limited resource use — deciding how much to take and where to leave alone. The haenyeo.

That structure is now collapsing. There were 23,081 haenyeo in 1965. At the end of 2025 there were 2,371 — down 252 in a single year. 1,500 of them, 63%, are over 70. Average annual income is 7.91 million won, less than half the minimum cost of living.

This is not a heritage-preservation problem. It is the collapse of the time axis — and observing a collapse in real time is data that cannot be obtained anywhere else, ever again.

The young who left, the retirees who arrived, the three-month resident. And memory that disappears unrecorded.

Jeju has a history of building graves constituted by memory alone. That domain belongs to the bereaved and to the islanders. It cannot be treated as material for a framework. One may learn from it; one may not absorb it.

Wind. Jeju's curtailment events rose from 3 in 2015 to 181 in 2023. In December 2024 the third Jeju–Wando interconnector opened, enabling reverse flow to the mainland; in 2025 roughly 11.6% of renewable generation went back that way.

The problem did not get solved. It moved.

If you can only export as much as your cable allows, which is the city-state's answer — more cable, or export without cable? Electricity that cannot travel can be converted into computation and sent anyway. That question is not Jeju's alone.

Hallasan, and what lies beneath it. The visible layer and the invisible layer are one system, but they must not share a name. The layer citizens walk into and the layer where machines turn must stay separate. Blur that line and a civilizational narrative becomes an industrial brochure.

Within the Sim Eternal City canon this already has a name. Prototype #2: Tamna. If the first prototype is a city newly attached beside a land city, the second is the case of an existing city read again.

A testbed requires three things. Jeju has all three.

Boundary. Experiments must be measured, and measurement requires a boundary. In mainland cities people and money cross freely, so effects cannot be isolated. An island is different: the points of entry and exit are countable. Jeju is one of the rare societies on earth large enough to matter and bounded enough to constitute a control group.

Scale. Too small and results do not transplant; too large and failure cannot be absorbed. A population in the 660,000s is large enough for significance and small enough to reverse.

Precedent. Tamna negotiated with the outside world under its own name 1,100 years ago. This matters not out of nostalgia. A society that has experienced self-definition changes frames faster than one that has not.

One thing must be stated plainly. This essay argues for neither secession nor independence. It is not a question of administrative status. It is a question of the unit of description.

When Jeju describes its own problems in the language of a city-state rather than the language of a province, its interlocutors change. When the interlocutors change, so does the menu of solutions.

The last point is the essential one.

A testbed requires someone commissioning the test. If Jeju experiments alone, that is a Jeju project. If six city-states test the problems they cannot solve at home in Jeju, that is infrastructure.

And the trade is not one-way.

The city-states hold what Jeju lacks — regulatory sandbox experience, digital citizenship regimes, capital, international protocols. Jeju holds what they lack — a closed boundary, early-onset data, the extreme conditions of a volcanic island and its sea, and a living case of long-validated self-governed resource discipline.

When both sides have something to trade, it is not aid. It is collaboration. At that moment Jeju stops being a place that needs help and becomes a place that is needed.

This is what was missing from the solutions I saw in Jeju. All of them faced inward. All of them asked what could be brought in from outside. None asked what could be given out.

And this is why a book written in Colombia is being read again in Jeju. The framework did not arrive to explain Jeju. Jeju is telling the framework what it left out.

Singapore · Tallinn · Dubai · Amsterdam · Nairobi · Tokyo

What six city-states would want tested in Jeju — and what they can put on the table in return.

Singapore has no room to fail. The entire territory is a system in production, so a city-scale experiment is a city-scale risk. Super-aging arrives in the 2030s, and by then there is no time to test.

Commission — an economy where the old remain producers. The haenyeo structure is the physical instance of what Singapore holds only as a blueprint.

Offer — city-state governance design. The full grammar Jeju needs to move from the language of autonomy to the language of a city-state.

Estonia is the only country that built a way to count citizens outside its territory. But e-Residency sits at the layer of tax and registration; it has never descended into actual community.

Commission — a structure in which those who leave remain citizens. If the roughly 2,000 twenty-somethings leaving each year have left only physically, how does a system that keeps counting them actually work?

Offer — a finished stack for digital identity and remote civic participation. Jeju need not build from zero.

Dubai has already finished the law. Law No. 9 of 2023 set the basis; the 2026 executive regulations fixed six vehicle categories, eight licensing conditions, ten technical and safety standards, and fourteen operator obligations. Vehicle approval and business approval are separate gates, and civil liability sits squarely with the operator. Commercial driverless service began in March 2026; since July it has been free to the public in Jumeirah and Umm Suqeim. The target is 25% of all trips by 2030.

The problem is that the data is all one kind. Desert, flat ground, new-build, young population.

Commission — volcanic terrain, high wind, ocean, and an aged population: the exact inverse. The Dubai Autonomous Zone is a 15 km corridor; Jeju could be a corridor the size of an island.

Offer — sandbox law as a built artifact. And something more important: Dubai proved that the city that writes the rules first gets the testbed. That is why firms take technology blocked at home to Dubai. It is also exactly the position Jeju should be competing for.

In 2021, following a citizens' initiative signed by over 30,000 people, Amsterdam set an annual cap of 20 million tourist overnight stays, with a clause requiring corrective measures within six months if 18 million was breached.

It has not held. 20.7 million in 2023, 22.9 million in 2024, 23.7 million in 2025 — three consecutive years over, each a new record. This despite a hotel construction freeze, halved cruise calls, and Europe's highest tourist tax.

Residents sued. In February 2026 the city told the court that the 20 million figure was an administrative ambition, never a legally enforceable limit, and that the ordinance contains no enforceable results-based norm.

Commission — whether a volume cap can exist in an enforceable form at all. An island is one of the few urban units where entry points are countable.

Offer — one expensively learned lesson. Setting a ceiling and making a ceiling binding are entirely different pieces of work.

The inverse curve. Young people pour in, the grid does not reach, and so the city skipped a stage into off-grid power and mobile money.

Commission — the model for converting surplus power into computation and exporting it. This is precisely what Nairobi needs too: how to export energy without cable.

Offer — the methodology of the leapfrog. What Jeju needs to learn is not the advanced case. It is this one.

Japan's vacant homes reached 9.0 million in 2023, a vacancy rate of 13.8% — both record highs, and double the 4.48 million of 1993. But what follows is the interesting part. Nomura Research Institute had forecast that 13.8% outcome as 16.9%, and revised its 2033 projection down from 30.2% to 18.3%. The panic about the total was overstated.

The real problem is composition. Homes that are neither for rent, for sale, nor second homes — simply abandoned — number 3.85 million, and their share of all vacancies has climbed from 36.5% in 1978 to 42.7% in 2023. Meanwhile 70% of Tokyo's 896,500 vacant units are rental stock, and more than half of the 128,000 housing starts in 2023 were, again, rental.

Commission — what happens when a city stops building and makes re-occupation its first instrument. Tokyo keeps building while vacancies overflow. Nowhere on earth has broken that loop.

Offer — a thirty-year time series, and a record of forecasts that missed. Thirty years of data on what does not work is more expensive than five years of data on what does.

bcdW Current Weekend Edition publishes every Saturday. Sim Eternal City — future city storytelling on cities, climate, and aging. Saturday contributor: Paul.

One City. One Story. Many Views.