The Most Important Thing

"A Dog of Wu" is in here!
#shamelessplug

It seems that my story, A Dog of Wu (appearing in the March/April issue of F&SF) got a mention at Tangent Online: "The author successfully creates a complex future society," they say. I'll take it - especially since, if you think about it, creating a world in 8,000 words is damned hard.
Which is why you shouldn't think about it, if you're a writer. DON'T BUILD A WORLD.
Oh my God. What is he saying? Is he saying that everything should be set in the present, or some easily extrapolated near future? Is he-
Nope-nope-nope. Of course not.
But let's face it: You and I have both read a lot of stories - short, long, and medium - with meticulously worked-out worlds possessed of their own languages, religions, sorceries, technologies, geographies, and ecologies, that somehow just don't work. What makes some thorough, logical universes as dry as dust? What makes an insane, inconsistent universe (e.g. Roger Zelazny's Creatures of Light and Darkness) brilliant and utterly convincing?
No, really, I'm asking. I have no idea.
Okay, maybe I have one or two-

The first is character. [Well, duh.] If the characters aren't convincing and sympathetic, it doesn't matter how much you know about this world.

The second, I think, is a compelling "Theory of Operation." If you know how the MOST IMPORTANT THING in your story operates, and why, the rest will fall into place. That's why I say "Don't create a world; create a Theory of Operation for your Most Important Thing."
F'rinstance...in most fantasy worlds, the Most Important Thing is magic. If you know how your magic works, and why it works that way, your world is more than half-built. Take Ursula LeGuin's A Wizard of Earthsea. If you've read it, I guarantee you remember the rules of magic. There are only two, after all: The name is the thing. The world's balance can be disrupted by magic. And that's it. Simple...but almost everything emerges from those two principles.
In science fiction, that ToO/MIT is often expressed as What if? What if robots have humanoid

intelligence, but are bound by three unbreakable laws? What if there's a biowarfare agent that turns its victims into single-minded analysis machines, the ultimate grad students? What if one small group of people have enormously long lifespans that are purely the result of hereditary, and the rest of humanity finds out about it?
If you take a look at those stories, you'll see that the level of worldbuilding detail goes from minimal (Asimov's I, Robot) to enormous (Vinge's A Deepness in the Sky) but they are all successful, because the What If, the ToO, is simple and compelling.
In A Dog of Wu, I didn't create an entire world. I created just enough of a world by focusing on my What If - what if humans were bred like dogs? What would that look like? Why would it happen? How would it work? Who would do the breeding? The rest of the story depended on that. If I had tried to create that world from scratch, instead of deriving it from its most important thing, I would have gone crackers.
You can spend weeks or months or years coming up with languages, races, and geologic processes; but if they don't hold together, they don't matter. If you focus on your Most Important Thing(s), everything else is more likely to fall into place.
That's my theory, anyway. What's yours?