Felixg , your post tried to compress a full article into a single if longer forum post. Impressive and there is probably a lot to add.

Just some related points:
I only know some about standards for medical devices. There the requirement that are put on the software (and hardware) are high.
For control and aviation software the requirements can be even higher.
The standards are VERY strict about security, redundancy, reliability etc.

This makes the software development itself just a part of the total development process, and not the biggest.
Aside from development there is verification and validation and all that jazz that the quality department is so happy about.
Designing tests for required functionality, identifying risks and lots more.
And the tests are not started yet!
See how the tests for the software for the F-35 take longer and longer time.

To get a glimpse of how detailed it gets and how high the requirement are,
NehemiaG once posted here that he was surprized a place did not keep a detailed list
.of the age of drills used in maintainance

In that kind of environment there is no simple way to fix bugs or add features without often adding massive amount of work.
So changes are clamped together, justify the extra work, few and far between.

Also none of this is new. many methods and at least one new branch of math (operations Research) were invented during the second world war and space programs.

Sing while you may
- The Legendary Pink Dots