[jeremysmyth]

Google and Facebook, amongst others, have shown that the quick iteration strategy works, even for complex, critical systems.

Complex systems, yes. Critical... well that depends on how you define critical.

The whole "quick iteration strategy" idea depends, largely, on the idea of "fail fast" or "move fast and break things". This necessarily makes iterative software development into an exploratory discipline, poking at something that doesn't work with a stick so that you understand it better and then creating something that does work off the back of that.

You simply cannot "move fast and break things" in a field where your created artifacts are expected to last a long time; we'd live in a fragile world of prototypes and conservatism were that the case.

In pretty much every other field populated by people who call themselves engineers, there's a whole lot more thought and maths going into the design of constructed artifacts, fed by a whole lot more science, before doing the actual construction. This approach is not just passé, but anathema in the software world. We just don't do engineering the way the rest of the world does.