How to Write a Long Fantasy Series

This essay was originally posted on LiveJournal as the conclusion to my re-read and analysis of Robert Jordan’s Wheel of Time series. Although it references the Wheel of Time, it does so without significant spoilers, and the argument as a whole is not specific to the series. My purpose here is to talk about the specific challenges of writing a long epic fantasy series — here defining “long” as “more than a trilogy, and telling one ongoing story.” (So something like Mercedes Lackey’s Valdemar books wouldn’t count, since they’re a conglomeration of multiple trilogies.) My points probably also apply to non-fantasy series, but other genres are much less likely to attempt multi-volume epics on this scale, so I’m mostly speaking to my fellow fantasists.

I do not pretend this is in any way, shape, or form a recipe for commercial success with an epic fantasy series. After all, most of this is a checklist of errors I feel Jordan made, and you could paper the walls of Tor’s offices in fifty-dollar bills with the cash he made for them. Nor am I claiming artistic failure awaits if you fail to heed this advice; you might squeak through on luck, or just really good storytelling instinct. But I do feel that bearing these points in mind can help the would-be writer of an epic series avoid falling off some of the more common and perilous cliffs.

With all of that intro material out of the way, let’s get to it.


On the basis of my re-read, and comparing to other series that attempt similar tasks, I have come to believe there is a single, fundamental principle, underlying all the other points I’ll make throughout this post, which governs the author’s ability to keep the narrative from spinning wildly out of control, to the detriment of their story.

It’s simple:


Most of us, when we set out to write a novel, have at least a vague sense of how long it’s going to be. We can be off in that estimate — In Ashes Lie ran about thirty thousand words longer than I originally intended — but generally speaking, you know that you’re aiming for 60K or 100K or 200K, and you use that to guide a thousand decisions you make during the process. Should you introduce new subplots, or is it time to start tying things up? Does your protagonist’s next action need some complications in its path, or would it be better to just handle it offscreen and move on to more important things? Can you bring in a new character for this strand, or should you find a way to take care of things with the characters you already have? These are questions of pacing, and we’ll come back to that a bunch of times along the way. But you can’t gauge your pace when you don’t know how long the race will be: at best, you’ll end up going through the whole thing with a steady, slogging, workhorse pace that (to switch metaphors) loses all sense of dynamics.

Pick a structure, and stick to it.

By “a structure” I mostly mean “a set number of books,” though I allow that there might be other ways to conceive of it. J.K. Rowling knew the Harry Potter series would be seven books, and each book would span one academic year at Hogwarts (plus or minus a little time before or after). The actual size of those books varied wildly, and you can certainly make the argument that she would have benefited from tighter editing as the word-count ballooned. But does anybody think that situation would have been improved by her saying, “There’s an awful lot of stuff to deal with in book five; I think I should split it in two”? I doubt it. (The decision to split the final film was likely drive as much by financial aspirations as artistic, if not more so. And oy vey is that the case with the two Breaking Dawn movies. But by then the material was set; the end was in sight.)

I haven’t read Steven Erickson’s Malazan books, but I’m told he set out to write a ten-book series, and that’s what he delivered. And you know what? Based on what I’ve heard from readers, some of them thought it was great, some of them thought it was flawed, but none of them thought it was the trainwreck of apocalyptically bad pacing the Wheel of Time turned into. Whether or not you liked where the story was going, it was indubitably going somewhere, and at a reasonable clip.

A Song of Ice and Fire, by contrast, was supposed to be a trilogy. Then a quartet. Then a sextet. Then A Dance With Dragons got too long, so Martin split it and now the series is a septet. In a recent interview, he said it might run to eight books instead. Step by step, I can see him walking into the same swamp Jordan got lost in.

Tom Simon discusses this in his essai Zeno’s Mountains, wherein he cites David Eddings saying that a man who’s never walked a mile has no real sense of how far a mile is. Most of us learn how much Stuff goes into a novel by writing one; we learn how much Stuff goes into a trilogy by doing the same. How many of us ever write more than one seven- or nine- or ten-book series, though? Jordan never got a chance to learn from his first attempt and do better the second time. Martin likely won’t, either.

Simon says, “I do not know of any general solution to this problem; perhaps no general solution is possible.” I say there is a solution, and its name is Discipline.

As answers go, it isn’t perfect; keeping your series confined within its intended boundaries may result in a less satisfying arc for various plots than you would get if you let them stretch out to their fullest. But letting them stretch may very well be detrimental to other aspects of the story. Keep one eye always on the larger picture, and know what must be accomplished by the end of the current book for you to remain on schedule.

Doing so may require some ruthless editing. And it’s entirely possible that such editing won’t be in your best commercial interests: it costs time and effort, laid against the odds that allowing the story to sprawl will translate into more money for you and your publisher alike. From the standpoint of craft, though, rather than the bottom line:

Pick a structure, and stick to it.

Continuing onward from there, I have learned several other salutary lessons, most (if not all) of them standing on that structural foundation.

2. Control your points of view.

A friend of mine, in discussion regarding an epic fantasy series she’d like to write, proposed that this should be the number-one item on my list. I put it at number two because I believe structure is one of the major yardsticks by which the decision to add a new pov character should be measured.

I could point to any number of cautionary examples from the Wheel of Time (goddamed Vilnar Barada comes to mind, or Alteima), but I think it’s best to look at the moment where I first noticed Jordan going wrong. That would be the pov scene for Jaichim Carridin in The Shadow Rising, the fourth book of the series — the one where the branching nature of the story is at its strongest, right before passing from being a feature into being a nigh-fatal bug.

For those who aren’t familiar with the Wheel of Time, Carridin is a minor villain character who gets four pov scenes in the entire series. In this particular scene, we discover that he’s scheming with Liandrin (another minor villain; she gets four pov appearances, too) on behalf of one of the factions he serves, and with the King of Tarabon on behalf of a different faction. Which sounds good, except that the key word in that sentence is “discover” rather than “scheme” — relatively little action takes place. Most of Carridin’s 3,194 words are spent on him thinking about stuff: the current political situation in the city, the current political situation outside the city, the way his evil overlords have been slaughtering his family one member at a time to motivate him, etc.

Some of the information that appears in this scene also reaches us via different channels in the story. Other parts aren’t terribly relevant, because they don’t come to anything in the long run. Jordan could easily have cut this scene, and we would have lost very little of substance; the few salient details could have been brought in elsewhere, by other means.

But let’s pretend for a moment that the information here is actually vital. Does that justify spending time in the head of this minor villain?

No. Because here’s the thing: switching to Carridin is lazy. It’s the easiest way to tell us what the bad guys are doing — and I do mean “tell,” given that most of the scene is Carridin thinking rather than acting. Had Jordan restricted himself to a smaller set of pov characters, he would have been forced to arrange things so that his protagonists found out what Carridin was doing. In other words, they would have had to protag more. And that would have been a better story.

[Edit: see the end of this essay for something of a counter-argument to my original point.]

Every time you go to add a new point of view character, ask yourself whether it’s necessary, and then ask yourself again. Do we need to get this information directly, or see these events happen first-hand? Can you arrange for your existing protagonists to be there, or to find out about it by other means? Are you sure?

Given what I said above about sticking to your structure, there may indeed be times where it’s more word-efficient to jump to a new pov, rather than constructing a path by which your existing viewpoints can pick up the necessary threads. But be careful, because taking the lazy way out appears to be a slippery slope for authors. This page lists no less than sixty characters who get only a single pov scene each during the entirety of the Wheel of Time. Nineteen more get two apiece. Eleven get three, seven get four, and then the numbers start ticking upward faster, until our six primary characters have between fifty-seven and two hundred — just to give you an idea of scale.

If I am counting correctly, this series has ONE HUNDRED TWENTY-NINE POINT OF VIEW CHARACTERS.

That is absurd.

Martin is starting to have a similar problem, albeit on a smaller scale. He has thirty-one viewpoint characters so far, according to this page. Fifteen of those — nearly half! — have been introduced or received pov in the last two books, and most of them have only one or two chapters apiece per book, well below the usual average for this series. One character in A Feast for Crows died at the end of his sole chapter, whereupon pov transferred to one of the people he’d been traveling with. Why not give that person viewpoint to begin with? Why not spend the pages developing that character, instead of the one who won’t be with us for long?

John Scalzi once pointed out the inexorable consequence of multiple points of view on pacing, which authors of long epics would do well to bear in mind. If you have a 120K book and one pov character, that’s a hundred and twenty thousand words forwarding that character’s story. If you split it evenly between two characters, they get 60K apiece. Four characters, and now each of them has only 30K in which to move forward. Pretty soon, it feels like not very much is happening with any one of them.

Of course, you can mitigate this to some extent by having those characters interact, so that A’s story is progressing even while we’re in B’s head. But that brings us to our next point . . . .

3. Control your subplots.

Once you have multiple pov characters, it’s easy to let them wander off from one another and start doing different things. This isn’t inherently bad; if you want to write a long epic fantasy series, you’re going to need a high degree of complexity. But if you lose sight of your structure, you’re liable to also lose sight of how many subplots is too many, and which ones are taking too long to resolve.

There are two ways to fall off this particular cliff. One is that you know X is going on in Y part of the world, but you’re afraid it won’t seem reasonable if you spring it on your reader at the point where X begins to affect the rest of the plot. (Or you just think it’s too shiny not to show, or whatever.) So you decide you need to show X happening — and probably add a point of view to facilitate that. The other path starts with the point of view: having given a character pov rights, you feel consciously or subconsciously obligated to justify that decision. On a small scale, this leads to pointless crap like Vilnar Barada thinking about the girl he wants to marry; on a large scale, it leads to things like the Shaido Plot From Hell, which I am convinced was Jordan creating makework so that Perrin would have something to do, and also justifying Faile as an ongoing pov character.

It may annoy readers (especially when you do it badly), but I’ve come around to the philosophy that you shouldn’t be afraid to give one or more of your characters a sabbatical from the story. The example of Jordan doing this right is Perrin’s absence from The Fires of Heaven: Perrin had just won a great victory and settled into some necessary but unexciting work of consolidation, so it was a dandy time to step away and focus on other characters. The story would not have been improved by inventing a subplot to fill that gap. The example of Jordan doing it wrong is Mat’s absence from The Path of Daggers: Mat had just been trapped under a collapsing wall during the invasion of a city. It turns out nothing interesting had been going on with him during his book-long absence . . . but given where the story had left off with him, readers expected a great deal more, and didn’t get it. If you’re going to step away, choose the point at which something has wrapped up, not begun.

Making up subplots to keep a character busy is a cascading problem. The proliferating points of view created and/or abetted new plot complexity, which meant the central ropes of the narrative got stretched out farther than they were meant to go. You can’t shelve your main character for three books, though, so Rand — ostensibly the driving force of the whole shebang — didn’t have a lot to do for a while other than run around micro-managing the politics of several nations, creating a lot of material that didn’t really add all that much to the story. It did add words, though, which meant Jordan had to find something for Perrin to do while Rand was occupied, so Faile got kidnapped by the Shaido, and then next thing you know, you’ve created a monstrosity of a plotline that 80% of your readers will hate with the fire of a thousand suns, and oh by the way now you need to keep all those secondary characters busy, too, the ones who started this problem in the first place. It’s the principle of the Lowest Common Multiple, played out in narrative form: if one character is cycling at 13 rpm and another is at 20, you have to keep rolling until you hit 260 to get them both wrapping up at the same time. And that way lies the ever-expanding tale.

If you stick to your structure, you at least have a metric by which to gauge whether a subplot is worth the time it will take to cover it. Of course, most of us can’t really eyeball an idea and say “why yes, that’s fifteen thousand words’ worth of subplot” — would that we could! But this gets back to the “ruthless editing” I mentioned before. If it starts stretching out too far, find a way to accomplish the necessary elements more efficiently. If you can’t do that, cut the subplot. Yes, it may be shiny, but is it worth throwing off the balance of everything else in the story?

4. Centralize.

This is closely-enough related to the previous point that I almost folded it in there, but I think it deserves to be pulled out and looked at on its own.

A long series is going to have a certain amount of sprawl, which is both necessary and desirable. But keep an eye on how long it’s been since your major characters interacted with one another. In the Wheel of Time, the fourth book was the first one where the main protagonists didn’t all come together for the finale; not coincidentally, it’s also the last one where the story’s sprawl felt truly effective. Something like eight or nine books passed without Rand and Perrin seeing one another, or Perrin and Mat. There was a point in the story where Rand, Mat, Perrin, Egwene, and Elayne were all in different places doing different things, and had been for some time; that’s five major plots rolling without reference to one another, in addition to the countless minor plots. We may also consider that Martin’s story and pacing have begun to fall apart as he lets his characters separate further and further: when’s the last time you had any two of Arya, Jon, Bran, Sansa, Catelyn, and Tyrion in the same place at the same time? (Not to mention Daenerys, off on the other side of the planet this entire time, or the host of other pov characters Martin has begun to introduce.)

Remember Scalzi’s point above: the more you fragment the perspective, the less forward movement each one gets per book. Remember my corollary: you can mitigate that by having the viewpoints overlap. Apart from the simple mathematics of pacing, this helps deal with the subplot issue, because you can keep important characters in the narrative by having A work with B on whatever it is B’s doing. (Or oppose it, or interfere with it, or whatever.) And it will assist in maintaining your structure, because if Aragorn’s got to be at the Black Gates when Frodo arrives at Mount Doom, then you’ve got to get that Pelennor thing done on schedule, which means not letting the Paths of the Dead episode overstay its welcome.

(Note that I am NOT holding up Tolkien as a model for how to construct the kind of narrative I’m talking about here. His approach was to ignore half his story for half a book, which isn’t a tactic that will serve any modern author very well. But Lord of the Rings is familiar enough to serve as a useful example.)

So yes. By all means let your characters wander off and do their own thing . . . but not for too long. Bring them back together periodically, and look for ways to get multiple stones to work together on killing that bird.

5. The further you go, the less you have to show your math.

This is less tied into the structural base than the rest of my points; it’s more a simple matter of word bloat.

Early on in your story, it’s useful to show how your characters pull off their small accomplishments. It demonstrates their competence to us, if it’s something they’re supposed to be good at, or conversely shows them developing new skills, if they’ve been thrust into situations outside their usual depth. Or it establishes the realism of the world, or gives the reader information about a topic they may not know very well. All of that is perfectly fine.

But when you’re ten books into your series, you really don’t need to show the camp logistics of the army your hero has been in command of for the last four books. You don’t need to walk through every step of how the heroine, having attained her throne, arranges a meeting with some fellow sovereigns. You’ve already established that these are tasks well within their skill-set. We will not bat an eyelash if you go straight to the meeting, or have the army keep trucking along in good order. If you introduce some element that makes those tasks hard again, then by all means show how the new challenge is overcome — but even then, you’re allowed to only focus on the challenging part, and let the routine stuff go.

Because in theory, the further you go into your series, the more exciting the story should be. Tensions mount! We’re building toward the climax! Now is not the time to stop and do the simple math all over again. Think of it like a geometry proof: once you’ve proved the basic theorems, you’re allowed to just cite them and move on, rather than having to go through every step every time.

One of the corollaries to this is more debatable. Re-reading the Wheel of Time, I was struck by how many times the story explains Min’s visions; it felt unnecessarily repetitive to me. Arguably, however, that sort of repetition is necessary, because some readers may not have read the previous book in a long time, and may have forgotten who Min is and what she can do. (Or they may have picked up the third book without having read the first two, though I tend to be of the opinion that people who do that deserve what they get. I note that many series, including both the Wheel of Time and Harry Potter, eventually give up on holding people’s hands — it just takes a while.) This is more a matter of exposition than showing the narrative math, and I’ll allow that some amount of reinventing the wheel may be required. But keep an eye on it anyway, and try to keep it to a minimum.

There are many other things I could say about the flaws in the Wheel of Time, or in other long series. But these are the main points, the ones I think are universally applicable, rather than specific to a particular narrative — along with, of course, the basic lessons of good writing, like not using twenty words where five will do. A story’s quality depends heavily on its shape, on the timing of various twists and revelations, the pacing of its arcs and the rate at which the characters grow; and good shape rarely happens by accident, especially on a large scale. Ergo, I firmly believe that you need some fixed points by which to navigate during your journey. Know how many books you’re going to write, hammer in a couple of pegs to say that certain events will happen at certain points, and then hold to your course. If you stray from the path, you may never find your way out of the woods.

Rumor has it, of course, that Jordan was asked to stretch the series out, because it was making so much money. I have no idea if that’s true. But as I said at the start, my concern here is not the commercial success of a series; I’m addressing the story itself.

I’m speaking, mind you, as someone who has yet to write a series longer than four books (and those structured almost entirely as stand-alones). This is all based on my observations of other people’s efforts, not my own experience. But as I said to Tom Simon in the comments to “Zeno’s Mountains,” there’s not enough time in life to screw it up yourself for a dozen books, and then to do better afterward. If you want to write a long series and not have it collapse in the middle like a badly-made souffle, you have to learn from other people’s mistakes.


The above wound up generating some interesting discussion in a variety of places: comments on the original post, Twitter, etc. I wanted to come back to it long enough to highlight a few lengthier responses that I think make very good points.

The first comes from C.E. Petit at Scrivener’s Error; scroll down to the third bit to find his thoughts. I tend not to talk about “theme” because the word has been so badly treated by high school English classes, but his point is a sound one, and can provide guidance as to how the author might gauge whether their story has begun to grow out of control. Are you diluting your thematic message by adding in all these other subplots? Or, conversely, are you hammering your reader too energetically with that message, by playing through sixteen variations on the motif? (Which is not, of course, to say that the work will have only one thematic message, especially if it stretches to four books or more. But a central line is still vital.)

The second, or rather the second and third, is Patricia C. Wrede’s two-part response to my own argument, which digs further into the question of why authors fall into these traps, and what they can do about them. I want to say that she is 100% right about the arbitrariness of your opening structural decision: even if you base it around some kind of pattern (as she suggests in the second post), ultimately that’s a framework you then try to pour your story into, rather than a natural outgrowth of the story itself. You don’t set out to write seven books because that’s precisely how much character and plot and so on you have to tell; you write seven books because you decided to build each one thematically around the seven deadly sins or chronologically around the years Harry will be in school, and then you try to scale everything else to match.

Note that we do this all the time in fantasy: it’s called a trilogy. You sign a contract for three books, okay, and so you plan your story based around that arbitrary decision. I’d venture to say that the vast majority of series that are planned as trilogies end up as exactly that. There are exceptions (Terry Goodkind, as discussed in Zeno’s Mountains; George R.R. Martin; the Hitchhiker’s series), but it seems that most of us are capable of sticking to three books when that’s what we said we’d do. It’s only when we go beyond three that our control seems so liable to slip — because we have so few models for how to do it right, and because one more book is much less expansion when it’s ten instead of nine than when it’s four instead of three. And, maybe, because if you’re selling well enough for your publisher to support nine books, they’re eager for you to make it ten instead.

But we manage it with trilogies, and TV writers manage it almost without fail when they write shows with season-long arc plots. Absent the network jerking them around, they finish their story in twenty-two episodes of X minutes each, period, the end, no “please just one more ep” or “sorry, this one ran twelve minutes long.”

Is that kind of discipline detrimental to the story? Sure, sometimes. But so, manifestly, is allowing one’s discipline to falter. And I say — with the spotless virtue of an author who has never yet had a publisher throw stacks of money at her, begging for a bestselling series to continue — that I would rather make myself find a way to tell my story more efficiently, with fewer digressions and wasted words, and end it while people are still in love with the tale, than risk losing sight of the original vision in a swamp of less productive byways.

(“You either die a hero, or live long enough to see yourself become a villain.” Speaking of tales planned as trilogies, and delivered that way, and in my opinion all the better for it.)

It isn’t easy. As Wrede points out, it requires frequent check-ins with your plan, however you may have built said plan. It may require you to murder some very beloved darlings. But just as a sonnet’s structure can force you to make really good use of your fourteen allotted lines, so can a fixed length to your series.

Addendum to the addendum: reading this over again some years later, I found myself frowning slightly at my own arguments about giving viewpoint to minor characters (especially ones you’ll never see again) and using that space to develop them in ways that aren’t relevant to the central narrative.

I mostly stand by what I said there, because I’m mostly talking about mainstream, modern, western fantasy here. But, of course, that’s not the only way to do things. Literature is full of stories in which sprawling digressions are a key feature; speaking of those I’ve personally read and enjoyed, the Mahabharata comes to mind. And an author may choose to focus briefly on somebody you’ll never see again, their own personal concerns and hopes and fears, as a way of reminding the reader that the world is a big place and doesn’t revolve entirely around the protagonist(s) — a very worthy maneuver!

For me, it comes down to that “feature vs. bug” question. In the case of the Wheel of Time and A Song of Ice and Fire, I feel like this habit is distinctly a bug. It’s something that develops later in the narrative, rather than being integral from the start, and because of that, it contributes substantially to the feeling that the author is losing sight of what they promised readers at the outset. ASoIaF was always braided out of multiple viewpoints (WoT actually wasn’t, not until later), but those viewpoints are presented as the central, important ones, the ones Martin wants you to get invested in. Introducing not just a single new perspective but a bunch of them, some of whom die an eyeblink later, most of whom are off in their own corners doing their own things, runs a high risk of making the reader ask “why am I reading about these people instead of the main characters?” If it were just a brief scene making a thematic point, it might be fine, but when it’s three chapters of delaying action that fail to really change the trajectory of the narrative, I feel that’s a poor choice.

But: I do acknowledge that there is more than one way to do this. So just think carefully about what your readers expect, whether adding a new viewpoint and plotline helps satisfy that expectation, and if not, whether you have a good answer for why you want to do something different. You may! And if so, I wish you the best of success with it.