Great post. I really enjoyed it.
The idea in #2 is really important — “launch as few things as possible”. I agree about the cost of features — they become so heavy to support, with testing, and documentation, and everything else. The more features, the harder it is to get other, more important things done. . .
That said, if you become to afraid to try out new features, you could miss out on things. So, it might be better to say “spend time removing features that nobody wants”. The stupid little feature you add could provide learnings for the future.