The issue is simply that accurate estimation cannot be done for complex projects, yet people who aren't engineers and especially politicians cannot psychologically accept this, so insist on setting fixed size budgets (which then inevitably turn out to be insufficient).

A better approach would be to do what's done more in the tech industry - decide a project needs to be done, then allocate it a permanent percentage of total revenues with close supervision of the details, until it's completed.

> The issue is simply that accurate estimation cannot be done for complex projects

I do not think so. There are plenty of previous examples with enough similarities to estimate costs to +/- 20% at most, usually better.

A bigger problem may be that quoting wildly optimistic price is more likely to get a public project started, even though most parties understand that the actual budget and time will be very different. But as long as the project takes a long time those who gave original estimates / started execution have plenty of chances to move away. And then "well, when I was running it, things were different" is a decent defense.

Maybe I should have said "cannot be done reliably".

Even a very inaccurate and bad estimation process will yield correct estimates at least occasionally - stopped clocks and all that. But estimate-driven budget allocation is hardly useful if the result is wrong 50% of the time, and to me it feels like it's at least that unreliable. Although this story focuses on Germany, look at CrossRail in the UK for another pertinent example. It was "on time and on budget" right up until a few months before launch, when it suddenly wasn't on time or on budget anymore, and in fact, after churning through some CEOs, eventually they got one willing to admit the truth - that nobody knew how long it would take to complete. But by then they had started shutting down the project, e.g. by laying off their entire internal and external communications team. It'd have been far more effective if they'd admitted up front they didn't know exactly how long it'd take.

That's actually more like the approach taken in the past in the UK for huge infrastructure projects, or road and power station, sometimes via setting up a specific (nation or city) owned company that got funding rounds! Yes, projects turned out more expensive than thought quite often, but it didn't make headlines quite the same. It was managed to stay good value rather than expected to hit a precise penny-accurate budget.

I think the modern tendency to seek cheapest bid, rather than best value, makes it far worse. In trying to save money time and again they end up spending far more...