If you want to write science fiction, you have to get the science right.
David Gerrold wrote an article on his wall that dissects the above assertion. He does a pretty decent job of it, he is David Gerrold after all. His final point is correct in the article. We aren’t reading the story for the science, we are reading the story for the story itself. It is important that the narrative not distract you from itself by introducing problems that draw you out of the narrative. That is the sole restriction on telling the story well.
In the novel Dune, there is no need to explain where the oxygen comes from, if there is enough room left in the writing for the reader to fill that explanation in for themselves. I haven’t read Dune in quite awhile, but I recall a character (Kynes, I think it was) puzzling about where the oxygen comes from in one of the ancillary texts that came with the first novel. It is a mark of Frank Herbert’s genius that he invented the Butlerian Jihad to whisk away the questions of technology in the novel. That is the reason there were no computers in the universe of Dune. Religious zealotry, a quick and dirty way to write off any technology that near future readers might wonder about, and also the underlying theme of the novel itself.
Dune was not about the tech so why get distracted by it? Dune is 89% politics, and as a political treatise, it is almost unequalled in fiction. I would bet that Dune survives this period in time, if anything does.
I’m a serious Niven fan, but Neutron Star has quite a few science holes in it. They didn’t matter when he wrote it because we (the readers) didn’t know any better at the time. Also? I still call the overarching genre Speculative Fiction. It’s less cludgy than F&SF to say. If I mean fantasy I say fantasy. If I mean science fiction (like Larry Niven’s body of work) I say science fiction. But if you are talking about everything from Poul Anderson to Roger Zelazny, you are talking about speculative fiction.