At Boskone I was on a panel about long SF series (despite the fact that I have never written one). Fellow panelists were John Douglas (once an editor of mine), Rosemary Kirstein, and Alastair Reynolds.
To succeed, a long SF series has to keep showing you new facets of the world. Sure, people like to settle in with familiar characters, and, to some extent, relive past adventures. Still, a good SF series is more like a work of architecture, rather than a painting (I wish I'd thought of this while on the panel). You can't see it all from one vantage point. You have to move through it, and while you are seeing it from one angle, there are things you can't see, though you might remember them.
The totality of that integration is a genuine aesthetic pleasure, one that gets shared on a narrative basis with an integrated multivolume work like Lawrence Durrell's Alexandria Quartet. The first three books of the quartet show you the same time and place from different points of view, all of which are to some extent incompatible. You read them in order, so each one denounces and corrects the previous version. It's all about love, one says. No, you got it all wrong, Balthazar says in his "great interlinear". No, it was actually all about politics says the third. Then the fourth book goes on, accepting all that but never resolving the various strands. It is that last element that is not like a work of science fiction.
Science fiction books provide the basic equation that moving through plot means acquiring knowledge, and that the knowledge translates into the power to move to another level, where you may well acquire more knowledge. SF narratives tend to have that open-ended structure, as each answer entails further questions. That is one reason some personalities find it so compelling.
This penchant for systematic investigation does have its downsides. Systematizing science fiction authors often reach a state of hypersystematization, where, like conspiracy cranks, everything needs to be explained in terms of everything else. Isaac Asimov, for example, reached this phase late in his career, when he felt the need for a unified field theory of Isaac Asimov: he tried to make out that every single one of his novels was actually a facet of a single universe.
Larry Niven seemed to get trapped inside his own creation of Known Space, and has long depended on collaborators to help him find a way out. So science fiction is also the home of shared worlds, where collections of writers play in a single universe, sometimes created collaboratively, sometimes leased from a single writer, like Niven, who has grown tired of pushing narrative through the increasingly narrow holes left in his own creation.
And you do get clubs, cults, collections of obsessive fans. That's just the nature of the root psychology of our field. The prototype is probably fans of Sherlock Holmes, with their finicky attention to the Canon, and the contradictions in it. The commentary on Star Trek, Tolkein, etc., dwarf all other commentaries in literature. You could establish entire civilizations based on them.
So, in SF, long series are not just a lazy way of reusing a background that took a lot of work. They are a different literary experience, one that seems long, but is actually thick. I don't think I have the stamina for doing one, but admire those who can manage it. It is a different type of work.