One of the projects currently sitting on my desk and glaring at me in a state of disarray has this issue I've come to call the Problem of Arbitrary Construction. Since the story is almost pure fantasy, there are very few constraints on what's possible in it, and that right there is a problem. Stories driven by arbitrary conceits tend to lose the reader's engagement, because in a story where anything's possible, nothing matters very much.
A common way to deal with this is to lay down some key rules in, say, the first third of the story, and have everything else evolve out from that. Once people have some ideas of what's possible, and what's at stake, it gets easier to not only use those rules to generate an interesting story and suspenseful situations, but also to break those rules when the time is right. The point is to pave the way, so that the breaking of the rules doesn't feel like just another thing thrown in to make things "interesting".
Another thing I try not to drown in is the laying-down of the rules. Some fantasy books spend a dismaying amount of time preoccupied with how the rules work, to the point where they are less fiction and more dramatization of a tabletop RPG system. A few basic rules are fine; everything outside of that tends to deliver diminishing returns.
Still, I kept thinking about that word arbitrary, and what kind of epithet that is when applied to the way a story can unfold. "Arbitrary" isn't when we have multiple fantastic things in a story, but when there is no common thread between them, no unifying origin. It's when the author seems to be thinking, if one what-if in a story is good, then two must be really good, and five must be absolutely gonzo! (No, it isn't.)
For me the worst kind of arbitrary is the kind that pays no mind to the fact that a story tends to happen to people, and that the people in those stories tend to have the psychological attributes we associate to people generally. And if they don't, then you'd better take the time to lay out why and how. A society that has social norms that are fiercely contrary to human behavior for the sake of imposing controls can't just be presented as-is; you have to, again, pave the way — give us some idea of why these behaviors exist and who benefits from enforcing them.
But most of the arbitrary things I see in stories aren't like this. They're matters of convenience or laziness or sloppiness; they're things that happen because otherwise we don't have a story. Or they're a form of invention on the part of the author where the mere fact of the invention is more important than what it means or what it leads to, as per two grafs up.
A story needs to have rules if only as a way to allow the reader to understand the terms on which it is to be approached. We can't play a game called "basketball" unless we have a set of rules that are generally recognized as being those that describe the game "basketball". Otherwise, we get Calvinball.
How does all this apply back to what I'm doing? That's what I've been wrestling with, because the story in question is driven mainly by the power of the unconscious. It could very easily turn into a wholly arbitrary set of events, with me handwaving away any apparent lack of connection by saying "that's how the unconscious mind works". Even now, sitting here typing this, I have to laugh at the lameness of such a dodge. My mission, then, is to make this project not fall into this trap — to take what could be arbitrary and random at a glance, and give it a discipline and a direction that only it can have.