Find a big problem, identify a basis for its vector space, and build that basis.
My opinion is that all "small problems with simple solutions" that have become successful were actually cases of big problems masquerading as small ones.
>You need to start with anything, NOW. No excuses. //
So I've got a problem and an outline solution. I just need R&D capital and a few months in a workshop to create a prototype, get a patent and then take that looking for investments ...
How do I start "NOW" with what I have, no money, no time, no workshop.
Time I can probably manage a little of, but then I use HN for relaxation when I'm tired so it's not really productive time that I'm using up. The project is nothing without capital and a workshop; just dead.
Sure there's other things to do but then I'm giving up on the project and that would be falling to the point you attacked that to do X [this project] I need Y and Z.
My opinion is that all "small problems with simple solutions" that have become successful were actually cases of big problems masquerading as small ones.