Ask HN: How do you suss out very hard but solvable problems?

I spent some time today discussing Elon Musk, the old PayPal crew, and how to get a nose for very hard problems that can, nonetheless be solved (advancing the art and establishing or furthering a business) with an old technologist friend of mine, who's been writing a book. While we agreed that "intuition based on experience" is part of the answer; we also settled on a few rules - I hope y'all can add a few more or poke holes in these.

1) How long since the problem has been seriously revisited? If your queries about new ways are dismissed with facile contempt, but no details and no examples and no specific journal references then maybe nobody alive or active in the field was around when the issue or technology became "settled." When Elon entered the rocket business, it had been an Oligopoly for a long time; many problems, such as how to form aluminum into a fuel tank other than by milling a huge chunk, hadn't been revisited in a couple of decades and more.

2) You hear "If it were possible somebody would have done it already" repeatedly from the "experts." This means "Even though this is my field and it would be an obvious advance I've never spent a second thinking about this." Which is odd, and suggests 1) is true, at a minimum, or perhaps that prevalent unconscious assumptions that have never been explicitly stated or tested, have decided the issue, at a maximum.

3) The question is just outside the understood boundaries of any current common field of expertise. Experts, we know from research, wildly overestimate their competence when considering questions bordering on their genuine area of expertise or outside it.

4) The topic has become disreputable, for reasons that are more coincidental than evidential. A highly publicized accident, fraudulent research in a similar area (that actually provided no evidence one way or another, since it was mere fraud.) Etc.