Every single major piece of infrastructure that you depend on today has a number of embedded assumptions. We assume population will be at a certain level, or will rise or even fall over the course of the lifetime of a project. Within a certain range of figures, the project makes sense. Go outside that range and the project turns into a bad deal. But how would we know it?
Routinely we make big ticket bets far beyond our ability to realistically project the future. We assume certain items which vary over long periods of time are constants. We can't help it.
This is bad enough when we know that we are doing it but very often we don't know that we're doing it. A century ago the US bet that the population will grow enough and be rich enough that we'll be able to replace our water pipes. We're falling billions short and living on borrowed time with pipes that have long exceeded their design lifetime. We're still using wooden pipes to supply water in some areas but until they break it's an unknown vulnerability. Year after year other projects that were more visible beat out the meat and potatoes issue of replacing leaking pipes.
Because we don't routinely track assumptions of population, income, taxation, debt loads, among others for each infrastructure project, it's not immediately obvious what projects are in trouble when a municipality starts shrinking, and when things are likely to start visibly going downhill. So long as we don't visibly track this sort of thing in a way that is widely distributed, it will be an easy call for politicians to stretch out painting a bridge here, replacing a pipe or a wire there. So long as it's only done occasionally and then made up, maintenance delay is recoverable. But stretch out the delays for decades and lives are put at risk.