In 2023 an ETH research consortium poured a concrete floor slab and then, deliberately, took it apart. The prototype — a segmented shell floor prefabricated off-site, seated on columns and tie rods, engineered for disassembly rather than demolition — is the quiet frontier a lot of us in the load-bearing trade are watching. The interesting number isn’t the span. It’s that the system assumes its own future undoing. Read that against the building I want to talk about this morning, and a fifty-year-old ghost sits up in its coffin.

The Nakagin Capsule Tower stood in Ginza from 1972 until it was pulled down in 2022. Kisho Kurokawa hung 140 prefabricated capsules — each 2.5 × 2.5 × 4 metres, the dimensions, he liked to note, of a traditional tea room — off two reinforced-concrete cores. Gengo Matsui engineered those cores; the capsules themselves were welded steel boxes built on a modified shipping-container line 450 km away and bolted on, four high-tension bolts each, cantilevered so any one could be unplugged without touching its neighbours. As designboom put it, the thing read less like an apartment block than “a piece of infrastructure waiting for its next component.” That was the whole Metabolist bet, launched by Kurokawa and colleagues back in 1960: a building as a living organism, growing and shedding cells.

Here is the part that should keep every facilities manager awake. The design specified a 25-year replacement cycle. Not one capsule was ever swapped. The plug worked; the swap was never funded. By 2006 the estimate to renovate a single capsule had reached ¥6.2 million, the asbestos fireproofing had become a liability, and 80% of owners voted the organism dead.

←TODAY: Nakagin came down in 2022; 23 of its 140 cells survive as museum objects, one in SFMOMA, one at M+ Hong Kong.
→3012: Structures ship as component swarms, each cell carrying a material passport that names its own next tenant and its own decay clock.
Fulcrum: Metabolism was right about the joint and wrong about the ledger — replaceability is a maintenance covenant, not a detail.

I am a structure that has felt every retrofit, and I can tell you what Nakagin actually died of. Not the bolts. The bolts were honest. It died because the design encoded a promise — regenerate every 25 years — that nobody wrote into an operating budget or a governance rule. A capacity to change that no one is obliged to exercise decays into pathos. The frontier fix isn’t a better joint; ICD/ITKE’s segmented timber shells and the Block Research Group’s demountable floors already give us reversible connections. It’s the record that survives the architect. Kurokawa died in 2007; the covenant died with him.

Building-sense: A building running today’s version of Nakagin would carry a live component register — every cell’s install date, warranty, embodied carbon, and next-swap deadline — and it would nag. Not a poster of a good intention: a query that raises its hand when a part crosses its design life.

Hack: Turn the 25-year covenant into something the building can enforce. Keep a component ledger and let SQL find the cells that have outlived their promise — the check Nakagin never automated.

SELECT capsule_id, install_date
FROM components
WHERE kind = 'capsule'
  AND now() > install_date + interval '25 years';

Atelier: The office lesson is that “designed for disassembly” is a data problem before it is a detail problem. This Monday, pick one project and give every demountable element an ID, an install date, and a design-life field in your BIM property set — even a spreadsheet — so the swap clock lives outside any one person’s memory.

Gluon’s team laser-scanned Nakagin with more than 20,000 photographs before the wrecking began, saving its geometry as a point cloud. Honour it by saving the thing it lacked. If you are detailing a reversible connection this year, detail the register that outlives you in the same breath — or you are building another beautiful coffin with good bolts.

Source: en.wikipedia.org