I think that there are 2 foundationañ problems that make complex projects over budget: 1. Stakeholders are incentivized to negotiate down the budget, skewing the result. 2. Breaking down a big task into smaller ones comes with the risk of missing some activities, but adding a nob-required subtask is almost imposible, also skewing the results.

Estimates of big endevours must be made by comparison instead of breaking down the tasks to avoid the second issue. That's one of the little secrets that make poker planning succesfull.