This has a lot of Steven Brust in it, because a) I’m currently going through a Brust re-reading phase and b) he’s one of the few fantasy authors I know who’s done more than one non-linear fantasy. He’s certainly the only one I know of who’s done more than five of them.
Why do non-linear narratives at all? Because they can add to the story. And sometimes varying the way a story is told can drag an author out of a rut she’s fallen into, which is bound to happen if she writes every single story the same way. And since many fantasies are set in a created world, non-linear narratives can become a way to introduce the reader to the world without info-dumping. And, finally, there’s the sheer pleasure and delight of creating new stories to look at. That last part is probably my biggest motive for writing at all.

1) “This is a manuscript from another world…”

Fantasy worlds accepted as distinct from Earth without having to apologize for it are a fairly recent invention. Before that, fantasy authors often had to put the story in a framework that the reader could accept as originating from Earth but not part of it: something like a dream (Eddison’s The Worm Ouroborous), a vision (Pilgrim’s Progress)…
Or a manuscript.
It’s perfectly possible to use this idea boringly, as happens when some more modern fantasy authors present the story as a translated collection of documents and make no attempt to create more of a framework than that. For one thing, it could well suck out the suspense. If the character is telling her adventures from the perspective of her eightieth year, it’s a pretty sure bet that she survived those adventures. For another, some authors feel compelled to put in annoying stylistic devices, like quotation marks in random places, just to remind the readers that they’re “reading” and not in the world.
Why not have fun with this? Tolkien had one kind of fun, “translating” everyone’s names from other languages and holding faithful to the perils of translation, such as putting in long passages in other languages that were not Westron/English (mostly in Elvish). Another is to consider how this manuscript came to be put together at all. Who assembled it? Why? Why did they choose this format? And, most interesting of all, what did they choose to put in, and what did they leave out?
This device is the best format I know of to make points about silence, the power—and threat—of history, and, when you really get into it, how characters speak to the reader. If the “editor” regularly interrupts the story to tell the reader that it got racy and so she excised that part, she emerges as a distinct, if annoying, personality. If the author cuts out the “boring bits” and presents the “good parts,” he creates the impression of a much vaster structure than was originally there; William Goldman does this with The Princess Bride, successfully enough to have fooled many people into thinking he was only the editor, not the author, of the book.
Consider doing this. And then go study about printing, bookbinding, editing, and related arts. If nothing else, it might make you reflect on your own process of writing, and whose eyes you choose to tell the story from.
And when.

2) “There are other times than these.”

Linear narratives often start in one place and proceed sedately to the end, interrupted at most by flashbacks or complementary sub-plots distant in space but not in time.
It doesn’t have to be that way, particularly if you have a huge amount of history and backstory to get to.
Carol Berg (who is one of my five favorite fantasy authors, and deserves to be better-known) starts at one point in her heroine’s life in Son of Avonar. Then she goes into what I thought was a flashback, to talk about what happened earlier in the heroine’s life. Then, I thought it was a really extended flashback. I eventually realized what she was doing: interweaving multi-chapter chunks of narrative, letting the character tell her own story at different points in time, without infodumping. It’s a rather startling technique, but for a story that is essentially two, divided in time, yet connected, and with neither thick enough on their own to make a book, it works.
Steven Brust really likes to experiment with time. Taltos, which is chronologically the first book in the Vlad Taltos series but was written fourth, starts out near the end. Then it begins interweaving two stories, one the protagonist’s autobiography (to quote several review pages on Brust, the book is written in First Person Smartass), the other the “main” plot of that book. Each chapter hops forward to briefly feed the growing timeline near the end, then returns to the two main interweaving ones for the chapter body. By the time that the book nears the end, you know what’s going to happen—mostly—but it still has suspense, a very different kind than most narratives do.
There is absolutely no question that Brust does this on purpose, for fun and to see what will happen. The Taltos series has no internal chronology. The “second” book in the timeline, Dragon (published eighth), has sections that fall after the “third” book (published second),Yendi. And because the narrator has a rather screwed-up memory and cannot always be trusted…
Oops, that properly belongs under the next point.

3) “Form follows function.”

Non-linear narratives are the most subtle ways of expressing a theme possible, I think. If your audience notices it, it will be great and reinforce itself through the very way they read the book. If they don’t, then your book (as long as it isn’t an unreadable mess) will still work as a book, and there is no obnoxious theme in sight.
Take Guy Gavriel Kay’s Sarantine Mosaic duology, which are my favorite series—the second book, Lord of Emperors, being my favorite fantasy novel ever. They leap about place and time and character, told in a mixture of tight third-person and omniscient that Kay flows between effortlessly. It isn’t until the second book that you really realize what he’s doing, though. All the same characters (except for dead ones) appear again, and phrases and ideas that first showed up in one scene repeat in others, tying the disparate scenes together. The series itself is a mosaic, which is its central artistic symbol. And considering that its other major themes are immortality, the survival of art, and how lives intertwine and cross each other to leave ripples years later, the form drives the theme home, hard.
Brust’s pattern is harder to see, but that’s because his series isn’t complete yet. There will eventually—one can hope, at least—be nineteen books in the Vlad Taltos series: Taltos, one for each of the seventeen Houses in Brust’s Dragaera (see point 5), and the last book, supposedly to be called The Final Contract, since Vlad’s an assassin. There are currently nine published, not counting the odd but connected Brokedown Palace and the chronologically earlier series, the Khaavren Romances. Throughout the books, Brust scatters little hints about Vlad’s past and future life, which may or may not be revealed in the books to come, and hints, as well, that Vlad’s memory has been deeply and powerfully modified by persons or forces unknown.
When the series is finished, presumably, the readers will know more than they do now. That doesn’t mean that we can see everything that’s happening now.
To create a deep structure like this and carry it off through multiple books is a pinnacle of one particular kind of fantasy writing. It’s truly using the market’s tolerance for multiple books in a series as it should be used. I want to do it someday.

4) “History is words.”

And many more words than fantasy authors might think, especially if they’ve never studied history. There’s fiction, of course, which most fantasies take the form of, and poetry and songs, which many authors don’t hesitate to incorporate. But they often don’t take into account that there are certainly other forms of writing in their created world. To wit:

  • Personal letters.
  • Diaries/journals.
  • Posters and broadsides.
  • Newspapers (especially if the world is based mid-Victorian or later).
  • Pamphlets.
  • Edicts, decrees, and other government documents.
  • Laws.
  • Travel books.
  • Religious books.
  • Birth records, marriage records, and death records.
  • Inventories (especially for large events, like the catering and entertainment for a feast).
  • Teaching texts.
  • Texts assembled by that world’s historians.

And so on, and so on, and so on. A great deal of the material of history is not fiction, though historians can learn from studying fiction. It’s those materials that don’t have as much prestige, are small and go unnoticed, and yet contain vital information that fictional authors often don’t think to record, or change as it suits them.
If you want to show off the history and completeness of your world, you can do worse than incorporate some of these materials into the narrative. It’s your choice how to position and connect them. They make wonderful transitions, but you don’t have to write a narrative that’s linear even in that respect. They can stand in for scenes that are necessary but would be hard to write. They can carry the book in complementary ways to the main narrative. They can even serve as the main narrative, especially in the case of letters, diaries, and journals. It’s probably not a surprise that Steven Brust and Emma Bull have written a Victorian epistolary novel. (‘Epistolary’ is just a fancy word for “composed of letters.”)
There’s an art to this, of course. You have to learn the style of such documents, or come up with one, so that they don’t all sound the same as the narrative. You have to make them interesting and vital, or people won’t read them. You have to learn how they differ in things like the passage of time from ordinary straightforward storytelling. This is where reading older novels that use the form of, say, letters can help. They can give you good examples of what not to do and what to do, too. If you find yourself bored silly while reading them, then take notes on what bores you, and don’t do that.

5) “Play with the paratext.”

The paratext is the bits that are part of the book but outside the main narrative itself, such as chapter titles, author introductions, and so on—those bits that the characters in the story wouldn’t have any chance of knowing about. Some of them, of course, the author usually can’t control, like the cover painting. However, you can have lots of fun with the bits that are yours.
Brust sure does. He arranged his first book, Jhereg, in 17 chapters on purpose, since each of them had an epigraph corresponding to one of the Seventeen Houses in his world that he wanted to introduce. Then the second book became 17 on accident. After that, he designed all of them to have 17 chapters on purpose, except when they weren’t divided into chapters at all. Where necessary, he cheats to get around the restrictions by having non-numbered epilogues, preludes, introductions, and so on. The longer Khaavren books have numbers of chapters that are multiples of 17, such as 34.
It’s in one of the Khaavren books, Five Hundred Years After, that Brust goes all out and has the book’s narrator, Paarfi, a writer of “historical romances,” interview Brust as if they were two living, separate people. They get into an argument about many things, including paratextual issues like the initials ‘P. J. F.” that appear on the title pages of several Brust books. It’s hilarious, and not only because of the clash of styles. These tricks aren’t really tricks, after all, but a way to deepen the world and give the author a rare chance to interact with his characters and readers knowingly.

6) “See with many eyes, speak with many voices.”

I often complain about viewpoint. This is because a lot of beginning fantasy authors, including in first novels, don’t know how to handle it. It just drifts around like a balloon, and while this sometimes produces neat effects, I don’t think that’s deliberate. I would much rather read a narrative where the author was in control of the viewpoint, even though it might take several attempts for the reader to be in control of it.
I had to reread Orca, the “seventh” book in Brust’s Vlad series, several times before I thought I understood everything that was going on. The books are usually narrated by Vlad in First Person Smartass. This one is narrated by a friend of his, Kiera, also in first-person. But Kiera’s narrative sections alternate with narrative sections that are Vlad’s first-person story, as related by him to Kiera. In between are snatches of conversation that Kiera is having with a third character, Cawti, and the book ends and begins with letters from Kiera to Cawti. The reader also gets several shining clues that Kiera is not telling Cawti everything, even as she says she is. This leads to questions about what Kiera’s hiding from the reader.
I found this gloriously complicated, and perfectly justifiable. The book is about Kiera and Vlad attempting to unravel a bank scandal, and the convolutions they have to go through match the structure. It’s also vital that the reader be kept out of Vlad’s head for at least part of the story, as we see at the end. And there are things that even Vlad doesn’t know, which Brust reveals in the last lines of the book. It’s a virtuoso performance.
Most stories don’t call for a narrative structure quite that complicated. But they could be more complicated than they are, and to some benefit. Does one part of the story really need a first-person POV to work, and the other a third-person? You don’t need to create a framing narrative device like a manuscript if you’re not comfortable with it (and if it would beg the question of how the first-person narrator came by the third-person narrative), and you don’t need to tell the whole story in just one or the other. Interweave them, and see what happens.
Essentially, what the story wants, the story should get. If it needs a linear narrative, give it that. If it doesn’t, start experimenting.

7) “Oh, but we’re all mad here.”

I usually find stories that end in “And it was all a dream!” examples of the Grand Cop-Out, where the author can’t think of an ending and so comes up with something stupid. This is because most stories of that sort aren’t written like dreams. They’re too logical, straightforward, and linear. There’s not a sign that they’re dreams. So the author tacking on that ending is either not too bright, or didn’t realize that the storytelling should match the narrator’s mind.
And yes, that applies if the narrator is mad or dreaming. It’s a little harder, but it doesn’t excuse you the effort.
Got a raving, delirious narrator? He’s not going to tell his story the same way that a clear-headed one would. A dreaming narrator should not question the illogical shifts that the dream narrative takes. A narrative told by a madman should be different, either openly or subtly, depending on the kind of madness, from the story of your typical perfectly sane, if rather overwhelmed, hero.
A question I’ve been tossing around lately is that of a story narrated by a non-human mind that doesn’t think or tell stories like a human. (This is because non-humans, and in particular non-human ways of storytelling, are an utter obsession of mine). If an elf can be, mentally, in several places at once, how does he tell a story? I think I might try writing one that way.
And that’s the end of that set of rants. The next poll should be up in a little while.