I guess I count as a “user from other Stack Exchange sites”, since there are 32 sites where I've written more answers than here, and I've never written a short story in anger.
My feelings are mixed. On the one hand, there are many threads here that are both fun and interesting. It does help that this site matches my interests pretty well — I'm an avid SF reader, and the world building aspect (including but not limited to counterfactual “what if” aspects) is what drives me to SF more than “mundane” fiction.
On the other hand, when I see topics I know something about, I'm sometimes disappointed (and occasionally even appalled) by the lack of breadth exhibited by some contributions.
With questions about speculative fiction, I sometimes want to scream “go read the classics!”. And TVTropes should be mandatory reading before asking — with all the SF that's been written, good luck finding an original premise.
When it comes to questions about realistic worlds, I especially notice problems in topics related to computer science (my field of professional expertise) and some domains of history and sociology (where it seems that some people believe that the whole world is exactly like their neighborhood). That makes me wonder just how much of what I read on other topics (physics, biology, history of parts of the world I'm unfamiliar with, …) is outright wrong, but just plausible enough for me not to be able to tell.
This is not a surprising problem on a site with such breadth. I'm not sure if it can be alleviated by any policies. An attitude of rewarding well-referenced answers and penalizing answers that lack both sources and reasoning would help somewhat, but I wouldn't want to impose a SkepticsSkeptics-like level of reference requirements either. I've sometimes noticed poor answers, decided to post my own better-informed one, and never gotten around to it because I'd have to spend too much time finding the right references.