“He wondered if this was something new…”

Note: This post is the fiftieth installment in my author’s commentary for Eternal Empire, covering Chapter 49. You can read the previous installments here.

When you’re taking apart a story to see how it works—whether as a fan, a critic, or an aspiring artist yourself—it’s often necessary to distinguish between two different kinds of narrative elements. One consists of the standard components that appear in the story to satisfy our expectations about the genre: a whodunit, for example, usually has a detective, a killer, a cast of suspects, and a series of clues, and most mystery writers assume that these pieces will be present before they’ve even come up with a premise. The second kind of element is designed to address the unique problems that this story alone presents: the details of character, setting, and incident required to enable the specific plot developments or twists that the author has in mind. And you can’t tell the difference until you’ve read two or more stories in the same genre, just as a scientist might need to examine multiple specimens to distinguish between the traits common to an entire species and individual variation. If you’ve only read one gothic romance, for instance, you have no way of knowing how many of those stock building blocks, from the orphaned heroine to the lonely house on the moor, are there primarily because that’s how the genre has defined itself. And it takes yet another level of sophistication to recognize that most of these elements exist because Charlotte Brontë needed them to tell a particular story in Jane Eyre. You can’t see the pattern until you have at least two examples to compare.

The same principle holds true for criticism—the best way to think about a work of art, after all, is to compare it to something else—and for the creative process itself, particularly when the artist is consciously drawing on more than one previous model. A sequel, even one that departs from the first story in surprising ways, always involves some degree of reverse engineering: the artist looks back at the original and tries to pick out the most important features, which create a sense of continuity between the old and new stories. (This usually, but not necessarily, means keeping some of the same characters, but it can also mean reproducing entire plot points or intangible qualities of tone.) And a third installment can draw upon two different models, which can both confuse and clarify the issue. When you mentally superimpose two stories, it naturally emphasizes the places where they overlap, and for the first time, it becomes possible to figure out which plot elements are essential to the series, rather than incidental solutions to specific problems. This is why an ambitious third installment can serve as a direct continuation of the second while simultaneously circling back to the first. Films like The Dark Knight Rises or The Bourne Ultimatum can struggle to manage those competing chains of meaning, which is why the fourth installment in a series often strikes off in a totally new direction: it becomes mathematically impossible, in the narrative equivalent of Metcalfe’s Law, to keep those connections straight when four or more stories are involved.

But it’s often not until the third installment that we know what the series is really about, since it takes two stories to establish the pattern. When I was trying to figure out the plot of Eternal Empire, I was very mindful of the problem of honoring the precedent set by the first two novels while departing when necessary from the template they had established—if only because it gave me a way to structure a novel that I knew was going to be very complicated. As I wrote in a blog post back when the novel first came out: “In a sense, [Eternal Empire] ends up serving double duty: City of Exiles ends on a cliffhanger that the third novel needed to resolve, but it also reaches further back to the first installment, so the resolutions of these two books essentially unfold in parallel before converging at the very end.” These days, I’m less confident that I pulled this off successfully than I was when the book was released, but I know that I gave it my best shot. And even at the time, I realized, consciously or otherwise, that the best way to keep the plot from spiraling out of control was to copy the structural bones of the previous books wherever I could. As a result, each novel has three acts, cuts between three main characters, covers roughly the same amount of time, and uses a narrative funnel that accelerates the action until the last section encompasses the events of less than a day. These are all solid, reliable thriller techniques, but it wasn’t until I had two prior novels to analyze that I began to understand exactly what kind of story I was writing.

You can see this clearly in the sequence of events that culminates in Chapter 49 of Eternal Empire. Both The Icon Thief and City of Exiles contain a plot twist that occurs at approximately the same point in each novel, involving the sudden death of a supporting character at the hands of an unexpected killer. This isn’t exactly unprecedented in this kind of thriller, and if it works, it’s all in the execution—and I think it works pretty well in both cases. (In fact, the two twists are similar enough that I went to considerable lengths to disguise it. This why the scene in the latter novel takes place in a car, rather than indoors, and it’s told from the point of view of the killer, rather than the victim.) I wanted a big twist in the third novel at more or less the same spot, both because I thought a reader would want it and because it served a useful narrative purpose: it occurs at a point where the story needs a jolt of energy to carry it through the last hundred pages. Because I couldn’t pull the same exact trick yet again, I ended up inverting it: instead of an unexpected death, it would be an expected death that turned out to be something else, with the reveal that Tarkovsky had conspired with Ilya to fake his assassination. It was like taking a piece out of the template, turning it around, and inserting it into the same place again to see if it fit. And it did. But if I hadn’t had two previous novels to study, I might not have known that anything had to be there at all. And I’m a little relieved that this series is over, because I’m not sure I’d be able to do it again…

The case against convenience

Last week, I finally bought a MacBook Pro. It’s a slightly older model, since I wanted the optical drive and the ports that Apple is busy prying away from its current generation of devices, and though it isn’t as powerful under the hood as most of its younger cousins, it’s by any measure the nicest laptop I’ve ever owned. (For the last few years, I’ve been muddling through with a refurbished MacBook that literally disintegrated beneath my fingers as I used it: the screws came out of the case, the plastic buckled and warped, and I ended up keeping it together with packing tape and prayer. If this new computer self-destructs, I assume that it won’t be in such a dramatic fashion.) And while it might seem strange that I sprang for a relatively expensive art object from Apple shortly after my conversion to an Android phone, my favorite thing about this new arrangement is that I don’t need to worry about syncing a damned thing. For years, keeping my laptop and my phone synced up was a minor but real annoyance, particularly on a computer that seemed to audibly gasp for air whenever I connected it with my iPhone. Now that I don’t have that option, it feels weirdly liberating. My smartphone is off in its own little world, interacting happily with my personal data through Google Photos and other apps, while my laptop has access to the same information without any need to connect to my phone, physically or otherwise. Each has its own separate umbilicus linking it with the cloud—and never the twain shall meet.

And there’s something oddly comforting about relegating these devices to two separate spheres, as defined by their incompatible operating systems. I’ve spoken here before about Metcalfe’s Law, which is a way of thinking about the links between nodes in a telecommunications network: in theory, the more connections, the greater the total value. And while this may well be true of systems, like social media, in which each user occupies a single node, it’s a little different when you apply it to all the devices you own, since the complexity of overseeing those gadgets and their connections—which are entities in themselves—can quickly become overwhelming. Let’s say you have a laptop, a tablet, a smartphone. If each connects separately with the cloud, you’ve only got three connections to worry about, and you can allocate separate headspace to each one. But if they’re connected with each other as well as the cloud, the number of potential connections increases to six. This may not sound like much, although even two extra connections can grow burdensome if you’re dealing with them every day. But it’s even worse than that: the connections don’t run in parallel, but form a web, so that any modification you make to one invisibly affects all the others. If you’re anything like me, you’ve experienced the frustration of trying to customize the way you interact with one device, only to find that you’ve inadvertently changed the settings on another. The result is a mare’s nest of incompatible preferences that generate unpredictable interference patterns.

Segregating all the parts of your digital life from one another takes away much of that confusion: you don’t have to think about any of it if your computer and your phone don’t speak a common language. (They can each talk to the cloud, but not to each other, which provides all the connectivity you need while keeping the nodes at arm’s length.) But Apple and other tech companies seem determined to combine all of our devices into one terrifying hydra of information. One of the big selling points of the last few Mac OS X updates has been a feature ominously known as Continuity: you can start writing an email or editing a document on one device and pick it up on another, or use your laptop or tablet to make calls through your phone. This sounds like a nice feature in theory, but on closer scrutiny, it falls apart. The whole point of owning multiple devices is that each one is best suited for a certain kind of activity: I don’t want to edit a text document on my phone or make a call on my laptop if I can possibly avoid it. It might be nice to have the option of resuming on one device where you left off somewhere else, but in practice, most of us structure our routines so that we don’t have to worry about that: we can always save something and come back to it, and if we can’t, it implies that we’re enslaved to our work in a way that makes a mockery of any discussion of convenience. And retaining that option, in the rare cases when it’s really useful, involves tethering ourselves to a whole other system of logins, notifications, and switching stations that clutter up the ordinary tasks that don’t require that kind of connectivity.

Is the result “convenient?” Maybe for a user assembling such a system from scratch, like Adam naming the animals. But if you’re at all intelligent or thoughtful about how you work, you’ve naturally built up existing routines that work for you alone, using the tools that you have available. No solution designed for everybody is going to be perfect for any particular person, and in practice, the “continuity” that it promises is really a series of discontinuous interruptions, as you struggle to reconcile your work habits with the prepackaged solution that Apple provides. That search for idiosyncratic, practical, and provisional solutions for managing information and switching between different activities is central to all forms of work, creative and otherwise, and an imperfect solution that belongs to you—even if it involves rearranging your plans, heaven forbid, to suit whatever device happens to be accessible at the time—is likely to be more useful than whatever Apple has in mind. And treating the different parts of your digital life as essentially separate seems like a good first step. When we keep each device in its own little silo, we have a decent shot at figuring out an arrangement that suits each one individually, rather than wrestling with the octopus of connectivity. In the long run, any version of convenience that has been imposed from the outside isn’t convenient at all. And that’s the inconvenient truth.

Nice to meet you, Harwin—or is it Hullen?

A few years ago, I wrote a blog post in which I tried to use Metcalfe’s Law to explain why ensemble casts on television can be so useful. It’s a formula that states that the value of a social network, such as a telephone exchange, is proportional to the square of the number of connected individuals (or, more precisely, n(n-1)/2). Any network, whether it consists of users linked by computers or characters on a show like Mad Men, gains its power less from the individual units than through their interactions, and with every additional member, the number of potential connections grows exponentially. That’s why television relies so much on ensembles: once you’ve run for a season or two, you’re constantly in search of interesting pairings that haven’t yet been explored in stories. A show with three major cast members has only three possible combinations, but with six characters, the number rises to fifteen, and while some of these pairings can be more fruitful than others—Monica and Chandler are more fun than, say, Ross and Phoebe—the odds of finding something that works increase with the number of theoretical interactions. Or so you’d hope.

In practice, of course, that level of connective density can pose problems of its own, especially in written form. When you’re reading a novel for the first time, you’re engaging in a complicated set of mental adjustments, which aren’t any less impressive for being so routine. You’re learning the rules of the world that you’re entering, making decisions about how fully to commit to the logic of the story, and figuring out who the hell everybody is. And the more names you’re asked to process at once, the harder it can be to surrender. In the first couple of pages of the opening chapter of A Game of Thrones, for instance, we’re introduced in quick succession to Bran, Robb, Eddard, Theon, Jon Snow, Jory Cassel, Harwin, and Hullen, all without a lot of handholding. This can be part of the fun of epic fiction, but I don’t think I’m alone in saying that my head started to hurt a little with every new name. Reading ought to be a left-brained process, and it can be exhausting when the right hemisphere is tasked immediately with keeping track of a dozen characters and their various relationships. It starts to feel less like entertainment than bookkeeping, or like being whisked through a series of introductions at a party at which you forget each name as soon as you hear it.

And we’re talking about George R.R. Martin, an old pro who is consciously testing the limits of how much information a reader can handle. In the hands of lesser writers, the chore of keeping the players straight can sap all the pleasure from the opening pages. (I sometimes feel this way when reading a story in Analog, in which I’m thrown a bunch of new names at once while trying to figure out what planet I’m on.) And it’s good for a writer to develop the habit of easing the transition into the story’s world as much as possible, especially in those crucial early stages when the costs of putting a book down are much lower than those of continuing. It helps, for instance, to introduce one major character at a time; to include short descriptive tags whenever convenient (“Bran’s bastard brother Jon Snow”); to omit names for minor figures, like the innkeeper we see once and never meet again; to keep the names you do use distinctive; and to make sure, above all, that each relationship is relatively clear before moving on to the next. There are times, obviously, when a writer will want to withhold this kind of information for a legitimate reason. In general, though, the cost to narrative momentum is so great that the situations where this makes sense are rare.

It’s also worth noting how much easier it is to keep track of the characters I’ve listed above on the television version of Game of Thrones. On television, in film, or on the stage, handling a large cast is a easier, both because we can rely on an actor’s physical presence to distinguish him from others and because our eyes are better at processing material like this than our brains are alone. (It’s a little like the difference between playing mental chess and looking at the pieces on the board: it’s harder to play a reasonable game when you’re preoccupied with remembering where everything is.) In Which Lie Did I Tell?, William Goldman advises writers to never open a script with a courtroom scene, since the stage directions can quickly degenerate into a list of names—the judge, the defense team, the prosecuting attorney, the client, the witnesses. He goes on to say that it’s fine to open a movie this way, since we’ve got the faces of the actors to help keep it all straight, and finally concludes: “I guess what I’m trying to say is don’t ask the screenplay to do what it has trouble with. Information overload is one of those trouble spots.” And this applies as much to fiction as to screenwriting. Your characters may be vivid in your own imagination, but to the reader, they’re just names on the page. And you’ve got to proceed with care and consideration if you want to turn those names into people.

Three (or more) is a crowd

There are no rules in screenwriting, as we all know, but one of them is this: you must never ever open your first draft screenplay with a courtroom scene.

—William Goldman, Which Lie Did I Tell?

He’s right. At first, a courtroom scene might seem like a decent opening for a movie. It satisfies the crucial requirement, as laid out usefully by screenwriter Terry Rossio, that every scene in a script be built around a clearly identifiable situation—and there’s nothing more familiar than a courtroom. We know the location, the players, the rules of engagement, and as a result, it gives us a convenient vehicle for generating suspense or drama. The sticking point, the pitfall that makes it impossible to use this as an opening scene, is the huge cast it involves. As Goldman points out, starting a screenplay in court involves laying out multiple characters in quick succession, and after we’ve been introduced to “Melvin Marshall, a bulldog in the courtroom” and “the legendary Tommy ‘the Hat’ Marino” and “Judge Eric Wildenstein himself,” our eyes start to glaze over. In a movie, this kind of scene works fine—we can use the faces of the actors to tell them apart. But in a printed screenplay, or a novel, all these names just blur together. Prose fiction is good at a lot of things, but one of its weaker points, especially at the start of a story, is introducing a large cast in a short period of time without confusing or annoying the reader.

Most good authors seem to understand this, but it’s one of the most common mistakes I find in beginning fiction. When I was reading submissions for my college literary magazine, almost without exception, I’d read the first paragraph of a new story, pause, and then read it over again, because the author was introducing too much information at once. There’s the protagonist, Gerald, and his sister, Sarah, talking about a third person, Horatio, whom we haven’t met yet, and they’re in the kitchen and it’s somewhere in Delaware and maybe there’s some kind of a war, and although I’ve been given a lot of material, I don’t have a single narrative thread to follow. Readers can handle a lot of complexity, but not when it’s deployed in one big lump. And while this sort of problem is much less common in professional short stories that have gone through an editor or two, it’s surprisingly common in science fiction. A lot of the stories in Analog, for instance, begin with a page that makes my head hurt, as we’re introduced to an exotic setting and some advanced technology and a bunch of alien names, and while certain readers seem to enjoy the process of puzzling out what the story is trying to say, I’m not among them.

The best thing a writer can do is begin by focusing on a single character with a clearly defined objective, and then gradually expand the narrative from there. You can, if you like, give us two characters in conflict, but no more than that, at least not until we’ve been adequately grounded in the players we’ve seen so far. Three is definitely a crowd. While editing the sound for THX-1138, Walter Murch discovered that when two characters were walking on screen, he had to carefully sync the sound of their footsteps to the movement of their bodies, but when there were three or more, he could lay the footsteps in anywhere—it was impossible for the audience to match the sound of individual steps to what was on the screen. This made his job easier, but it also led him to conclude that audiences, in general, have trouble keeping track of more than three elements at once. And this applies to more than just sound. Metcalfe’s Law tells us that the value of a social network—like a cast of characters—is proportional to the square of the number of players, and while this complexity can be wonderful when it comes to the overall shape of a story, when presented to us all at once, our natural response is to become frustrated and bored. Presenting the characters one at a time, and giving them clear objectives, is the smartest way to avoid this.

And although movies and television are significantly better than prose fiction at presenting us with a large cast, the best of them approach the problem in the same way. As I’ve mentioned before, there’s no better introduction to an enormous cast than the opening scene of The Godfather, with does precisely what I’m advocating here: it starts with an extended close-up of a minor character, Amergio Bonasera, and allows him to fully explain his situation before cutting to Don Corleone’s response. Later, at the wedding, we’re introduced to each of the major characters in turn, and each is defined by a clear problem or objective. As the movie progresses, these characters will acquire staggering complexities—but it’s that first, simple introduction that locks each of them into place. A similar process occurs in the pilot for Cheers, in which the regular characters enter one at a time until the show’s world is fully populated. By establishing the characters gradually and clarifying their relationships one by one, you’ll prepare the reader or the audience for the complications to follow. Once all the characters have been introduced, you can take full advantage of the possibilities that a large ensemble presents. But don’t do it all at once.

Bridesmaids, Metcalfe’s Law, and the power of ensembles

On Friday, my wife and I finally caught Bridesmaids, which is a classic example of energy and a star-making performance (by the sensational Kristen Wiig) bringing out the best in a formulaic, if nimble, script. It also benefits, like most films from the Judd Apatow factory, from a remarkably deep bench of supporting actors, including Maya Rudolph, Rose Byrne, Ellie Kemper, Jill Clayburgh, and Jon Hamm. The ensemble is so good, in fact, and has the potential to pair off its actors in so many surprising ways, that it’s something of a disappointment when the movie starts to focus exclusively on Wiig. We’re given a couple of scenes with the bridal party as a whole, but they all occur in the movie’s first half, and we’re never given the sort of inspired, inexorable comic set piece that the chemistry of the cast might have led us to expect. (Perhaps that will have to wait for the inevitable sequel.)

The movie’s decision to shy away from its supporting cast—the characters played by Ellie Kemper and Wendi McLendon-Covey, in particular, all but disappear in the third act—is a puzzling one, both because of the thrust of the marketing and because ensembles, especially in comedy, can result in unforgettable moments. Many of the recent films in the Apatow universe have revolved around putting a bunch of funny actors onscreen, rolling a lot of film, and hoping that something great happens. And occasionally it does. This is especially true of in television: even a mediocre episode of The Office, for instance, is usually worth watching for the sake of the cast, which retains a lot of viewer goodwill and still yields unexpected combinations. And as I’ve said before, it was Mad Men that opened my eyes to the potential of large casts of characters and the possibilities they provide.

Ensembles are particularly useful in television, where the various arrangements of characters can supply material, hopefully, for years of stories. To put it in the nerdiest terms possible, it’s an instance of Metcalfe’s Law, which states that the value of a social network is proportional to the square of connected users (n², or, more precisely, n(n − 1)/2).  A cast of characters is a peculiar kind of social network: it’s assembled by a producer, set into motion by the actors and writing staff, and its value lies in its connections, as various characters collide in interesting ways. The number of dramatically useful interactions also tends to increase over time, which is why the second and third seasons of a good television show are often the most interesting, once actors have had a chance to discover their most fruitful combinations. (Which is also why it’s sad that so many promising shows never get the chance to find this rhythm.)

Of course, there are limitations to such a model. Too many characters, and the show may never get the chance to adequately establish its supporting cast, so the pairings seem forced or arbitrary. (See: Glee.) But if exercised judiciously, it’s a useful tool for all kinds of narrative fiction, including the novel—and particularly for writers who otherwise tend to overlook such possibilities. As I’ve mentioned in previous postings, my first novel was a fairly focused story, with a limited number of important characters, largely because the plot itself was already so complicated. The sequel has a much larger cast, partly because I wanted to put some of Mad Men‘s lessons to use, and because I hoped that an expansive supporting cast would take me to interesting places. And I’m not the only writer to recognize this. In one of the notebooks he kept while writing Doctor Faustus, Thomas Mann writes:

Nothing yet has been done about staffing the book with meaningful subsidiary figures. In The Magic Mountain these were provided by the personnel of the sanatorium, in Joseph by the Bible; there it was a question of realizing the potentialities of the Biblical figures…The characters will have to be supplied out of the past, out of memory, pictures, intuition. But the entourage must first be invented and fixed…

More than almost anything else, a rich entourage of characters, if it arises naturally from the plot and setting, can take the story in unexpected directions. A large cast isn’t always a good thing. But if you’re looking to expand the world you’ve created, there’s no better way than to select two characters at random, put them in a room, and see what they have to say.