Sorry, Richard. But you always seem to settle for the trivials.

For trivial problems, you can create a design document, then implement, test and release.

What you never, ever, seem to consider is the projects that are a large percentage research. You can write a one-page document describing what you hope to manage to solve.

But then, you need to spend days, weeks, months, doing research. And lots of that research may actually involve writing real code. Then you measure, evaluate and rewrite.

So in many situations, you can't do that documentation up-front.

If I want a better melon, I can't write a 100-page document describing it. Then just implement, taste and accept. I need to spend months or years to perfect that melon. I can write a diary about the progress. But have to wait for the last part of the project before actually documenting the process that did give that better melon.

Some project can be: take 1 square meter of board space and compress into 200 cm2 of board space by using better technology. In that case, you can write a full documentation for the task since you have to know what the huge original design did. But that is only a limited part of all development actually made.