I think the bigger thing is that we think about requirements, but they change so frequently they require rework. I was working on an FMS for a customer and they changed their high level requirements every 3 months or so, driving huge costs and timeline overruns. It's easy when the Air Force gives you a solid set of requirements to work with, but what happens when the Army 5 months later gives you something contradicting those?

The problem isn't that we don't think about requirements. The problem is that there are no requirements. The Bomb and the SR71 had two very clear things in common - they were built to address existential threats. The bomb was designed for one reason - to fight Nazis we knew they were building the bomb and we needed our own. We knew exactly what to build and we went and figured out how. The SR71 exactly the same issue - we knew exactly what we wanted, we wanted a super fast undetectable spy plane to spy on the Communists.

The problem today is that we've got no damn clue who we're meant to be fighting against and it's been 30 years since the US has engaged in a war where air superiority was even an issue. The level of development in the US right now isn't "We need to build a system that's better than North Korea's" it's more like "Hey guys, I totally think we could build a system that plucks NK's ICBMs out of the air before they even jettison their first shroud".

There's no reason clear and direct motivation for these systems. So the purpose gets lost and we build something with a thousand bells and whistles because we're desperately trying to find some reason for it to exist.