Like I said - a proof of concept is only a POC if major features and plumbing are just missing.

You only build the parts you need to show the concept is possible. The bits you don't know exactly how to do already.

If it can be pushed to production usefully, it's too full-featured to call a POC, in my view.

That's exactly why I work like I do, now.

Unfortunately, I can't do it when I'm working for anyone else. They never let me do it the way that I do it. I have to be doing it on my own.

It works. It really works, if I write top-quality code, the whole time; even if I have to throw away or drastically refactor/write production-quality code.

My experience is that the final product:

1) Is of almost jaw-dropping quality (it was tested thoroughly, the entire time),

2) Is tailored exactly to the needs of the end-user (which deviated substantially from their original requirements), and

3) Arrives very, very quickly (this always surprises people).