This analogy emerged out of a lunch-time discussion too. The analogy goes like this "requirements analysis is like painting the Sydney Harbour Bridge". Sounds cute, but what does it really mean?

Well, painting the Sydney Harbour Bridge takes a long time, in fact no sooner have they finished they need to go back to the beginning and start all over again - I'm sure other bridges have similar problems or worse.

The point is that when you have a scope that is so large it is effectively impossible to survey all the requirements and expect them to be stable for the duration of the analysis phase. The larger the scope, the harder it gets.

I guess this is why I am skeptical about project methodologies that advocate nailing everything down to the nth degree before getting started on getting some value back to the customer. Methodologies that seem to work are the ones accept this volatility and set some broad visions but try to constrain detailed requirements analysis until just before implementation.

The vision needs to be defined just enough to get some unity across the deliverables but still allow some wriggle room.

Just-In-Time-Requirements-Analysis (JITRA) means that end dates need to be a bit rubbery which some people have trouble accepting, but my argument goes something along the lines of "the project team is going to work as hard as possible, but no harder, which means that the project will be delivered as soon as possible, but no sooner".