The wonderful SF Signal is back with another Mind Meld. This one asks, "Do science fiction authors have an obligation to be scientifically accurate with their stories? Is there a minimum level of accuracy an author should adhere to?" Responses include those of Pyr authors David Louis Edelman, Alexis Glynn Latner, and Adam Roberts, though my favorite points are raised by Karl Schroeder and Elizabeth Bear.
Schroeder turns the question around, arguing that science itself progresses by looking for holes in the contemporary understanding of the universe, and thus, "If scientists are obligated to look for holes in the 'scientifically accurate' picture of the world, would it make sense for science fiction writers to be obligated to uphold that picture?" (Which is an excellent anti-Mundane argument.)
Bear says, "None whatsoever. With one notable exception, which is to say, when writing rigorous quote unquote hard science fiction. I do think the SF writer has an obligation to know which rules she's breaking, and break them for a purpose, as an author writing historical fiction should alter history with intent rather than from ignorance."
The always erudite Adam Roberts reinforces Schroeder's point when he cites Paul Feyerabend's Against Methodand says, "Scientific rules limit possible advances in science: the only principle that does not inhibit progress is: anything goes."