Every water management intervention inherits a prior decision. Before the embankment was designed, before the drainage system was planned, before the early warning network was installed, someone decided that this was a place where people would live and invest. That decision is rarely examined by the water management community, which arrives to find a population in place and treats its location as fixed. See The Safety Paradox for what happens when protection makes a bad location feel safe. See The Productivity Trap for what happens when investment reorganises a community around its own continuation.

The most expensive water management problems are not engineering problems. They are location problems — and no engineering solution exists for a location that was always wrong.

The Netherlands is the most elaborated example in the world of a society that decided, over centuries, to inhabit a location that was marginal by any natural measure. A third of the country sits below sea level. The land was claimed from the sea through extraordinary feats of engineering — drainage, pumping, embanking, and dredging on a scale that transformed a delta into one of the most densely populated and economically productive regions on earth. The engineering was genuinely remarkable. And it worked — well enough, and long enough, that the question of whether the location was sound stopped being asked. The protection made the wrong address feel like the right one, and the development that followed made retreat progressively harder to contemplate.

Sea level rise is now forcing the question back into the open. The Dutch water management community is beginning to use the language of managed retreat — acknowledging, carefully and incrementally, that some parts of the country may not be defensible indefinitely, and that the honest planning question is not whether to retreat but when, from where, and on whose terms. It is one of the most politically difficult conversations in Dutch public life, precisely because the sunk cost of three centuries of investment in the wrong address is so enormous that acknowledging it feels like a betrayal of everything that was built.

The same pattern appears wherever settlements have been established in floodplains because the land was cheap and the flood had not yet happened. Wherever coastal communities have grown behind protection works that made a hazardous shoreline feel permanent. Wherever mountain villages have expanded into avalanche paths because a generation passed without an event. In each case the location decision preceded the water management challenge by decades or generations, and the water management community inherited it without questioning it.

At the sharp end of the pattern sits Blatten, a village in the Swiss Alps destroyed in May 2025 by a rock-ice avalanche of exceptional scale. Authorities had monitored accelerating slope movements for weeks. Early warning worked. Lives were saved. But the village was gone, and no engineering structure could have prevented it. The mountain was always going to move. What Blatten makes undeniable — in a way that slower processes allow to be deferred — is that the prior question is not how to protect a settlement in that location. It is whether a settlement in that location was ever defensible.

The engineering response has a cost that compounds over time. Each intervention that makes a bad location viable attracts more investment, more population, and more political commitment to its continued viability. The wrong address becomes harder to leave with every year of successful protection. And when the protection eventually fails, the losses are catastrophic precisely because the engineering response delayed the reckoning without resolving it.

The honest alternative is not abandonment. It is a prior question, asked before the engineering begins: is this location viable under the full range of conditions it will face, including the conditions that no protection can address? Where the answer is no, the conversation must shift — toward managed retreat, toward relocation incentives, toward insurance and early warning as the primary tools, toward accepting that some places cannot be made safe and that saying so is itself a form of water management.

Before designing a protection or management intervention, ask the prior question — is this location viable, and was the decision to settle here sound? Where the answer is no, say so clearly and shift the conversation toward managed retreat, relocation, and risk transfer rather than engineering solutions that delay the reckoning without resolving it. Where the answer is uncertain, make the residual risk explicit and build it into every planning and investment decision that follows. Do not allow the sunk cost of existing settlement to substitute for an honest assessment of whether the location can be defended — and at what price, and for how long.

Linked patterns: The Safety Paradox — protection makes the wrong address feel like the right one, attracting the investment that makes eventual failure more catastrophic. The Productivity Trap — investment reorganises communities around their location in ways that make retreat progressively harder. Privatised Gains Socialised Losses — the profits of developing a hazardous location are private; the costs of protecting and eventually abandoning it are public. Relief Crowds Out Prevention — the disaster that vindicates the wrong address generates the funding that rebuilds it in place.