AV Club: You’ve described much of the interaction in the show in terms of games. Is that how you see human interaction, moves and counter-moves?

Vince Gilligan: Well, I don’t know if I see human interaction that way in real life. In the writers’ room, our responsibility is to tell a good, interesting dramatic story. To be showmen, as it were. And to that end, we’re doing our best to come up with scenes that are as dramatic as possible. With all of that in mind, I suppose the best way to derive these moments of drama and produce them is to think in terms of gamesmanship.

Last week, football returned. 28.5 million fans watched the first Sunday's marquee matchup between the San Francisco 49ers and the Green Bay Packers. The collisions were real -- including a controversial flying lariat aimed at the 49ers' quarterback as he passed out of bounds -- but the game was fiction in every other sense: numerous storylines contributed to the game's allure (the 49ers ousted the Packers from last year's playoffs, San Francisco is a title contender this year) and new ones issued from it (the late-hitting linebacker had to defend his reputation). And of course all sports are fundamentally theatrical:

Ladies and gentlemen, the entertainment tonight will have four acts, with a twenty-minute intermission, played in an amphitheater. This piece of grass will be worth points, that piece will be out of bounds. Whoever has the most imaginary points at the end, wins!

28.5 million people devoted three hours and twenty-two minutes of their Sunday to this. And why wouldn't they? Games of all kinds make for great drama. Smart writers can exploit this.

In this piece we'll discuss the two games that fiction plays, and how we can get better at playing them. This essay is long (10,200 words), so before we lay the first bricks, we'll need some intellectual scaffolding. What should we call these two games? How do they inform each other? What are their characteristics?

Agon

The primary game is played between the characters. In A Song of Ice and Fire, for instance, this is the game of thrones. The teams are the various dynasties, the object is the Iron Throne, and there are no rules: backstab, poison, lie -- whatever gets you power. I'm going to crib a term from Roger Caillois and call this sort of game an agon. It's an ancient Greek word meaning "contest" or "competition", and applied to athletic and dramatic contests. It also gave us the words protagonist and antagonist.

No matter the particular type of agon -- the game of thrones, the drug game, the game of love, war, politics, life -- we love stories about games and competition. Consider the gamey aspects of America's favorite fictions -- top-grossing films.

• Harry Potter -- there are four houses/teams. They actually score points. There are wizarding tournaments, descriptions of magical sports. And all war stories are games. (The reverse is true, too... football people love to make their game out to be a war.)
• Hunger Games -- Obvious.
• Lord of the Rings -- The ring is a football. Mordor's the endzone.
• Twilight -- Team Edward vs. Team Jacob. Any love triangle is a game.
• Titanic -- See above.
• The Avengers -- A team of superheroes coached by Nick Fury.
• The Dark Knight -- A game for the soul of Harvey Dent, who likes to flip a little coin.
• Star Wars -- A walk on jedi hits a game-winning shot on the daunting goalie known as the Death Star.

I could go on: The Saw franchise, Battle Royale, The Running Man, Rat Race, etc.

This is not to say that every story's plot is or should be game-based. The point is that readers are definitely receptive to agonistic narratives, provided it's a good game.

What makes for a good game? Late in the fourth quarter of thrilling games, you'll hear a lot of play-by-play announcers say something like, "You couldn't script this!" But that's exactly what we're attempting -- so how would you do it?

You probably have a rough idea: pit great, colorful players against each other; ensure an even matchup, with lots of back and forth; keep the outcome in doubt until the last moment; maybe let the underdog win? By the end of this study we'll understand why these work, and have even more moves at our disposal.

The Metagame

While the agon is played by the characters, the reader and writer play a second game. In it, the reader is a gambler betting on the story's outcomes, the odds of which are set by the writer, or bookie. (Bookie is short for bookmaker, fittingly.) We will call this the metagame, since it takes the outcomes of the agon for its inputs.

Though the attitude of the players in the metagame is ostensibly antagonistic -- fans of Joss Whedon, George RR Martin, or Vince Gilligan will wryly damn these writers for their enormities, then wait eagerly for the next one -- the goals are ultimately cooperative: reader and writer both want a good story. This constrains the writer. You must "beat" the reader, but not so badly they want to quit. It's a delicate challenge familiar to any casino owner: the house must always win, but it should extract the gambler's cash gracefully, perhaps comping some free drinks in the process. Ideally, the swindled gambler walks out feeling that their time was well spent, if not their money. Here's a testimonial from a satisfied customer of Breaking Bad, John Lopez at Grantland: "That’s how you end a story. Vince and Friends played me like a broken pinball machine and, man, did I love it."

It shouldn't be so hard to play the reader, since the writer rigs the outcomes of the agon, and therefore completely controls the metagame. But readers have their own counterintelligence techniques that we will discuss later. For now, I'd like to illustrate the interplay of the agon and the metagame through a micro-story:

There once was a man, Bernard. He bought a new pair of shoes, and as he walked out of the store, another man stepped on Bernard's foot.

What happens next? It's hard to say, since this agon is quite bland. It's not much of a game at all, really: who's playing, what is the score? Let's add a few words so the metagame can be played.

There once was a man named Bernard. He was vain, and had a temper which got him fired from his last job. Depressed by this long stretch of unemployment, Bernard felt his luck was turning when he got a date with a woman named Cindy. To look his best, Bernard bought a terribly expensive pair of suede shoes. When he walked out of the store and headed to the restaurant where he was meeting Cindy, a man with muddy shoes stepped on Bernard's foot.

Easier to predict, right? It's likely Bernard is going to blow up at this clodhopper, maybe even start a fight. You can predict this because you have learned the character's temperament, which predisposes him to freak out in this scenario. That information is valuable both to the agon and the metagame. The fact of his vanity may serve the plot later in some other context, while in this moment affecting your perception of the metagame. Your confidence in a freakout depends on how you evaluate the various information: how vain is he? How bad is his temper? How hard is it to get mud out of suede shoes? You weigh these knowns against the relevant unknowns, like: how depressed is Bernard feeling today? How big is Bernard? How big is the guy who stepped on his foot? The writer controls all these variables of the agon, and needs to know how they will sway the reader's certainty. Once they all point the reader in a certain direction, you take the story the other way, and put one over on the reader. Or not, depending on how satisfying the obvious move would be.

There's your core loop. The agon generates the metagame, which may in turn affect the agon. To perform well in one, you must be skilled at the other. Even a really competent dramatist is going to bore people if he's got no tricks up his sleeve; a brilliant meta writer won't have a chance to bamboozle the reader if the agon's not exciting. With this system defined, we can now talk about how to manage it.

Building a better agon

Let's tackle the agon first. What are the components of a competitive game?

• Players
• Goal
• Object (sometimes)
• Score
• Rules
• A field of play

Players

In agonistic stories, our interest defaults to the characters that affect the game. Within that subset of characters, we only care about their personalities to the extent that they affect the players' performance. It's like a sports team: you pay more attention to the starters than the benchwarmers, and the stars' personalities don't matter until they affect their play. Derrick Rose really likes candy; I hear he got a Skittles machine installed in his house. That does not affect his pick and roll play. Ron Artest has a volatile temper: that does matter, since he's constantly getting into situations a more phlegmatic player (Tim Duncan) would not.

(And if you were the commissioner, you'd rather have a league full of Tim Duncans, tremendous players with flat affect, than a league of Artests, good players who might get ejected at any moment. We'll see the danger of incompetent characters shortly.)

It's easy to map this logic onto fiction. What do you know about Walter White, really? He's an excellent chemist, he got screwed a long time ago in his professional life, he's tremendously proud, he got a cancer diagnosis, he's married, has kids, he used to teach, and his brother-in-law is a LEO. That's not a long list for five seasons' worth of television. Think about all the stuff that never comes up: what are his parents like? What was his childhood like? How'd he meet Skyler? What exactly happened with Gray Matter? He has almost no backstory, and that's fine, because we know everything that is relevant to the game.

And note that Walt is the Tim Duncan of meth cooking. I used to think that having a "five-tool" protagonist was a power fantasy for both the writer and reader. Wouldn't it be nice if we were so good at anything?

Then I wrote an agonistic story about a bench warmer. That was deliberate: The Chosen One was my way of kicking the tires on the fantasy genre. And so the protagonist, Bryn, has no particular abilities, or at least none relevant to the war he finds himself in. Because he can't compete along with the other players, he's relegated to the bench, far from the fighting. I was concerned about how this would go over with the reader.

Look at audience backlash against Skyler White and Betty Draper. Leaving aside the sexism of it, I think there's the sense that these women are obstructing the action. Breaking Bad and Mad Men's main agons both have to do with work, and since Skyler and Betty aren't involved with that work -- and frequently are trying to pull their men away from that work -- they come off as momentum killers. This is a difficult problem to solve for the writers: these women are both narratively indispensable -- without a family to provide for, Walt could simply surrender to cancer; without a wife, Don's infidelities do not exist -- and yet their narrative task is a thankless one. One solution is to find an agon for them to take part in, ideally the story's primary. (Skyler's a lot more vital when she's aware of Walt's work.) Another would be to make them more appealing. (Betty's had a renaissance on Mad Men, partly because she no longer is fighting to hold a marriage together, and can be cool, cutting, and blasé towards Don.) The third would simply be to treat them more seriously as subjects, and not just motivations in the protagonist's life.

I tried the third way with Bryn: I hoped he would be interesting regardless of what he was doing. The psychological portrait I drew of him was not, as it might seem, my attempt to smuggle a literary character study inside an adventure story. It was intended to establish a rooting interest. To return to the hoops analogy -- we don't form opinions about players simply based on their skill. We like some guys because they play for our hometown team, or because they seem cool in interviews, or do funny dances on the bench.

Now Bryn wasn't meant to be a fan favorite, but I did intend him to be intriguing. He's a blend of privilege, self-loathing, a sheltered upbringing, and a strong nuturing instinct. For most of the book he does little to impact the agon, but all of this character development helps keep him relevant until he suddenly does get into the mix, and makes his contributions all the more meaningful once he does.

Before that point, though, I felt my interest was drawn magnetically towards the characters that could do stuff. My particular favorites were Edgar and Barabo. Edgar is the king's spy, which means he can act as liaison between the world of the palace and the rebels. No one else can cross the lines like this. In light of his unique role, he's integral to the plot's function, and therefore got more pagetime. I found myself giving him more and more character development whenever he popped up. It occurred to me, very late in the writing process, that I could have very easily made him the protagonist, instead of Bryn.

Barabo was similarly useful to the plot. He's as pure an antagonist as I've got in the story, and the best instigator. Stuff never failed to ensue when I brought him out of the wings. He's not terribly complex: there's little to learn about him beyond his attitude and penchant for knives. But that's enough. If you're interested in the narrative value of flat characters, check out the Jaime Lannister piece and CTRL+F for Joffrey.

Not every character has to be a dominant player, though. There's value in having lesser players -- for one thing, this is realistic, and for another, their weakness highlights the all-star's skill. But there is a threshold for incompetence: it seems that even the agon's losers must be minimally skilled. Those that aren't provoke the readers' scorn.

Prometheus, the Ridley Scott movie, got dinged on this account. There's a scene midway through the movie where a geologist and biologist get lost in an alien facility. Because of a dust storm, home base can't retrieve them: they must spend the night in the haunted house. To pass the time, the biologist hotboxes his space suit. During the night, a pale serpentine alien emerges from the black muck surrounding the scientists. And here's where the backseat drivers go nuts. Faced with an alien life form, the biologist gets curious. He extends a hand to it. Starts talking as if it were a stray dog. From the backseat: "You idiot! It's going to eat your face! Run!" He doesn't, not even when its cobra-like "hood" extends. The backseat: "Of course! Classic threat posture! Any biologist worth his salt would know this!"

Better creature design could have obviated this criticism. No one takes issue with the great Jurassic Park scene in which Wayne Knight runs into the "spitter." That's because the dinosaur, which also has a hood, appears much less dangerous than the alien snake: at first it coos.

The other fix would be better character design. Readers will grudgingly accept behavior they consider stupid if it has been set up beforehand. Had the biologist shown great curiosity before, perhaps the snake encounter would have played better. Or if he was simply a dolt; Prometheus had no outs here, though, since this is a billion dollar expedition. Therefore it is inconceivable to the backseat readers that the characters would be anything but hyper-competent all the time, since they're the best that money can buy.

The takeaway is this: your agon must be scripted so as to pass muster with the armchair generals. Bone-headed decisions must be justified somehow: the readers want high-IQ players. (IQ being used in the athletic sense, here.) You're going to get crushed if a character makes a bad decision because of insufficiently detailed emotional causes, or because of stress. (Though this may be mitigated if the result is sufficiently awesome; remember that the reader, like you, still prefers a good story over a correct story.)

Goal

I've been using sports examples for simplicity's sake: those games are much less complex, and much more explicitly defined, than those played in fiction. Consider something as basic as a cops 'n robbers agon. (We'll use Breaking Bad once again.) Clearly Walt, the crook, and Hank, the cop, are participating in a game together, but only incidentally. They are pulled into this game by their different goals.

Hank's goal as law enforcement is to uphold the law. Walt's goal is to make millions for his family, and so he's competing with anybody else in his line of work. Left at that level of abstraction, there's no reason he should be on Hanks' radar. But since his method of making money happens to be illegal, Walt is forced into a kind of competition with Hank. This is a little unusual, and worth pointing out, because Walt's proper competition is other drug lords. And so we shouldn't conceptualize all agons as boxing matches or something like it, with your protagonist in the blue trunks and the antagonist in the red. In fact, in the world of Breaking Bad, Hank is the referee in the bow tie, trying to make sure nobody breaks the rules.

So most agons are composites arisen from the collision of smaller agons. This emergence is fascinating to me, and managing the interplay between them probably merits an article of its own. Here's a synopsis of what that future article might look like:

1. Characters participate in multiple games simultaneously. Walt is playing the family game, the drug game, and the morality game, all at once, and so much tension emerges from it. In one game, he wants to be a good person. In the other, he has a problem which would be solved with some terrible behavior. Which does he prefer at any given moment?

2. Players in one game become objects in another. From a cop's perspective, criminals are prey to be hunted. From the criminal's perspective, cops are defenders keeping them from the end zone.

3. Some games have asymmetric sides. A Song of Ice and Fire is symmetric, like a marathon. There are a ton of competitors all vying for the same goal, The Iron Throne, and they're generally using the same techniques to go after it. War stories are more like soccer, where you've got offense and defense, and which goal you're pursuing depends on context, and it switches all the time.

4. Characters' goals or relationships may shift over the course of the story.

Are certain goals better for an agon than others, as certain characters are? No: any goal works, so long as it's supported by the rest of the text. I can be just as gripped by a person trying to find a job as I am by a superhero trying to save the world. To make sure a goal is well supported, and something the reader would liked to see achieved, here are some techniques:

• Have an active opposition defending that goal.
• Give a character multiple goals, whose fulfillments are all mutually exclusive.
• Make it tough to do (a hole in one is much more remarkable than a goal in soccer, which is much more remarkable than a goal in basketball).
• Make the reader feel that the consequences of a loss would be particularly painful. (Save the world narratives screw this up all the time. We have seen nobody else in the world. As far as we're concerned, the death of humanity just means the death of the six characters we've met.)

Object

Some sports use objects, like a ball, to define the action. And what is the venerable McGuffin but a football? It's uninteresting in its own right, but without it, passing the goal line means nothing, and so the players battle over it. Not because of what it is, but because of what it enables.

Other sports use objects to modify the players' abilities. The modern tennis racket has pushed the action back towards the baselines. NASCAR racers depend on their cars. There's been some noise about banning the belly putter in golf.

Agons can take advantage of both of these. In Lord of the Rings, the ring is both a football -- the endzone being Mordor -- and a nice little power up.

Don't forget that people can be objects, too. Anyone without agency qualifies, like young children: what is the purpose of the baby in Breaking Bad if not to be wrestled over?

All damsels in distress are objects because they, like footballs, go only where taken by the agon's subjects. Once put in the tower cell by the ogre, they stay there until the hero opens the door. And like a football, the damsel is not desirable for her own sake, but because she pads protagonist's box score. This objectification can disturbing, of course -- people aren't footballs. Decide for yourself how this kind of mechanic aligns with your politics; there's no need to include objects in your agon. But why do so many stories objectify in this way? What is the purpose of the object, and what makes a good McGuffin?

Objects focus and incite action. An object's value depends on how many of the characters desire it, and how badly. Just like with the goals, the reader must be sold on the importance of acquiring the McGuffin. Lord of the Rings spends a lot of time talking about just how badly Sauron wants the ring, and just how bad it'd be if he got the ring. Further, it shows us how powerful Sauron was when he had it. The pursuit of the ring gives us the suffocating scenes with the Ringwraiths, but also inspires Boromir's betrayal, and Gollum's involvement. No wonder it gets billed in the title.

Other thoughts on successful objects:

• The more purposes an object has, the more kinds of characters will pursue it.
• Objects that empower their bearer can allow you to convert weak characters into strong characters. This is the best of both worlds, some might say: you get a character with the relateability of an everyman and the power of a superman.
• Objects can be exchanged, stolen, gifted, destroyed, and transported.
• Objects can warp the agon more than any character, and yet do not require characterization. This allows more room for character development.

Score

How do we keep score, as readers?

It's tough: fiction lacks the unambiguous point-scoring of sports. There's no agreed-upon definition of a point, and certainly no video review or even boundaries to indicate when a point has been scored. In fact, in a two-person scene, there may be three different scores: one for each character, and then the reader's. This is a good thing, of course; it's perversely satisfying to see a character convinced he or she is winning. Since the score is tied to goals, all I'll add is that the reader has to know who is winning and who is losing.

Rules

Even if the nihilists have it right, and everything is ultimately without meaning or value... you don't want to invite one to game night. ("The only charade I see here is human existence, Laura.") For games, agons, and societies to function, the participants must suspend their disbelief and act as though it all matters. The rules in particular: we can't award a triple to a baseball player who exits left from the batter's box. Some actions must be permitted, and others proscribed -- otherwise you've got chaos. Taken as a body, these rules describe the game, and set up a structure to contain meaning and eject meaninglessness. If a ruleset seems limited, that's good: constraints create focus.

The rules for sports mostly dictate how the athlete's body can be used. In our dramatic agons, competitors use their minds as well as their bodies. Therefore, a more comprehensive set of rules is necessary.

When you think of rules for human activity, you think of the law. But agons also include informal rules such as social mores and folkways. If you've ever wondered what makes for good world-building, this is it. Not the nonsense census data you've compiled on continents which have never existed, but rather the behaviors which govern human interaction in that society. Steven Pressfield wrote a novel about Thermopylae called The Gates of Fire. I knew a little about ancient Greece when I read it, and emailed him to ask why he glossed over the pederasty endemic to that society. He explained in his reply that he considered it "vampire video", like half-naked girls in a music video -- something that would prove so distracting to his readership that it would suck all their attention away from the themes he wanted to address. I was somewhat disappointed by this ahistorical treatment, but he was absolutely right: could you imagine asking a reader to root for a protagonist who acts like Jerry Sandusky?

If our most serious taboos are or have been quotidian to various societies, any rule is at some level arbitrary. Our cultures seek to naturalize them and in the process obscure this fact. In this way, social mores can assert the authority of physical laws, like gravity; when this act of legitimization fails, penalties are assessed. These serve as deterrents, and also great engines for drama: rules were made to be broken, after all.

What's more, we learn about characters by which rules they observe and which they don't, since rules define behavior, and behavior defines identity. Batman's refusal to kill separates him from the criminals he fights. That code of conduct defines him as a vigilante with limits. For civilian readers, that's great. We like to see justice served, and are assured by the narrative that the goons he haymakers richly deserve their broken jaws. But in some Batman stories, the Law's institutional aspect decries Batman as a criminal. This seems unfair to the reader, of course: he's on your side! But the police's goal is to uphold the law, and the law and justice do not correspond exactly. Since Batman's ruleset includes violence committed without the state's imprimatur, he frequently breaks the law. Thus, characters with similar goals can be brought into conflict through their opposed methods.

When you impose rules on a personality disinclined to follow them, a great character can result. I adore both iterations of Tim Olyphant's Furious Lawman: Seth Bullock in Deadwood, and Raylan Givens in Justified. Both are lawmen with illegal impulses. In one episode, the bloodsucking father of Bullock's lover comes to camp. And so Bullock, the Sheriff, beats the guy's teeth in for the crime of being a creep. Watching Bullock negotiate that conundrum -- I want to punch this guy in the face but I can't -- is reliably entertaining. It's also a little sub-game that we understand very well. Whenever Raylan Givens encounters some criminality, we get to place our bets: is he going to go by the book? Or is he about to shoot this guy just for pissing him off?

Someone like Raylan has chosen to internalize a ruleset, but other characters find themselves forced to abide by rules they see no value in. Mad Men has two agons: one that takes place in the office, and another that takes place in the bedroom. These frequently clash; the whole show may be understood as a study in what happens when you mix business with pleasure. The work world has different rules than our social world, and playing by one world's rules while in the other leads to trouble. Don breaks rules in both spheres, and if you've watched the show, a highlight reel of his "greatest hits" is now running through your head. I'd like to draw attention to one that might not pop to mind right away.

In season two, episode four, Don and Betty are arguing about the kids: Betty feels that she is outnumbered, and that Don doesn't help when he comes home. Within this domestic agon, Betty's goals -- to raise the kids and to keep her sanity -- lead her to demand that Don make an effort. We suppose that raising the kids does fall somewhere in the list of Don's goals, but it's way down there. Top of the list? Keep own sanity and Don't get hassled. So Don threatens her: "You want me to bring home what I got at the office today? I'll put you through that window." Already this is horrible, but these two have argued before, and Don says nasty stuff all the time. Betty, furious at this threat, shoves him. And Don shoves her back.

I remember being astounded by this at the time. Don, as a husband, is garbage. He's unfaithful, deceptive, mentally and verbally abusive. But he's never gotten physical. The shock in Betty's face confirms this; her shove was predicated on the belief that Don would respect the rules of the non-violent husband. But in that moment, Don's desire to not get hassled by Betty was so great that he started playing by different rules.

So you can see how the rules we observe change as our goals do, and our goals fluctuate hour to hour.

Stray thoughts on good rules:

• Codes of conduct can shape relationships in the same way that a common goal or common enemy might.
• To break a rule in a satisfying way, you should first demonstrate its durability. If Ghandi kills somebody, boy that guy must have had it coming.
• Play up the conflict between rules and goals whenever possible. Cf. every Superman story ever.
• Different rulesets apply in different contexts. You can do things in wartime that would get you locked up back home. As characters move between contexts, they must deactivate conditioned responses, or be forced to play by rules they don't like. (The veteran must restrain his anger when a barfly berates him, for instance. And what would the barfly do if dropped into a warzone? You can't trashtalk your way around Kandahar.)
• Every person is assigned various rules based on their gender, culture, occupation, personality, social status, politics, etc. All of these circumscribe a person's possible actions, as they'd like to satisfy every ruleset at once. It's interesting when that is impossible.

A field of play

Boardwalk Empire does not care as much about the Atlantic City boardwalk as its title might suggest. The show doesn't contort plotlines just to keep a character on the boardwalk. It follows characters, even when they scatter to locations as farflung as Cicero, Illinois, or some cold farm in Wisconsin. I would admire that more in a different kind of show. Make no mistake, Boardwalk Empire is an agonistic story. It just spends too much time following the players, even after they've left the court, the locker room, and the stadium. An agon this diffuse has a hard time reacting.

At the same time, we should also acknowledge the value of multiple environments within a story, all segregated from each other. These barriers keep certain mixtures from taking place until the appropriate time, and bound the possible actions. Which, as mentioned in the rules section, is good for business.

• Pressure builds in tight spaces. The bottle episode isn't just a good idea for overbudget TV shows. Again, constraints work.
• In ecology, environments with regular -- but not too frequent -- disturbances harbor the most biodiversity.

Game Changers

Now that we've outlined the ideal components of an agon, how should they interact? A great matchup alone doesn't make for a classic game: what is the activity that creates drama in an agon?

In other words: what are game changers?

I hope you'll excuse the simplicity of this deduction, but:

The game changes when its components change. In sports, this usually means one of the following: a player is knocked out of the game with an injury, there's a major swing in the score, the ball is turned over, a bench player comes in and ignites a big run.

Agons can easily reproduce these moments: a character can be removed from the agon. A character can have the rug pulled out from under them just as their victory seemed assured. The McGuffin changes hands. A new character enters the story.

Since agons are more complex than sports, there are other possibilities. A character switches sides. A character embraces a new set of goals, or starts acting according to a new set of rules. Perhaps an object in the story starts acting like a subject.

And of course, the corollary is that these game changing events are multiplied by the significance of the player affected. The death of a minor character is, by definition, a minor death. The death of the major antagonist causes a much greater shake up. The boundaries of the field changes, or maybe it becomes a different field entirely. An agon is concluded, and a new one begins.

Below is an incomplete list of moments that can disturb an agon. There will be some repeats from the above paragraphs and lists, since I anticipate using this as a cheat sheet later:

• Players
  • A relationship changes.
  • An alliance is formed/broken.
• Goal
  • Character embraces a new goal.
  • A goalie skates into a previously open net.
  • A goal becomes impossible, perhaps destroyed. / A new goal reveals itself to the characters.
  • Previously secure resource is jeopardized, forcing an emergency pursuit of goal X.
• Object
  • Object changes hands.
  • Object is destroyed.
  • Object is revealed to have more power/different power than was previously thought.
  • Object is created.
• Score
  • The weak becomes powerful / the powerful is laid low
• Rules
  • A character begins playing by different rules.
  • A character breaks one of their rules.
  • The rules change (new regime, new environment, etc.)
• A field of play
  • A character dies/leaves/is disqualified from the agon.
  • A character appears/is born/somehow enters the agon
  • The characters enter a new environment
  • An environment is destroyed / discovered.

And here's something we haven't mentioned before: agons are full of hidden information. So until the characters realize the agon has changed, it hasn't. (In an infidelity scenario, for instance, you can have two agons overlaid while you wait for the cuckolded spouse to figure it out.) You can exploit this by keeping the readers in the dark, too.

If you have a good memory, you may recall somewhere at the beginning of this giant essay a mention of the metagame. We'll deal with this now.

As I said before, the metagame is a game of prediction and deception. At the end of every scene, there's an implicit question posed to the reader: "So what do you think happens next?" Readers love to be stumped by this question. The sensation of total surprise/suspense/possibility is what many readers chase in their fiction, and if a writer can deliver it, the job is done. (Well, almost. We're completely leaving out aesthetics in this piece.) Imagine you wrote Star Wars. How gratifying would it be to watch this clip? Look at this kid's face! But you know what's cool? We don't grow out of that kind of reaction. To wit: Game of Thrones reactions. (That's a spoiler for the end of season 3, so if you haven't watched it, don't click.)

How do we craft those kinds of moments for our readers? To answer that, we have to talk about the entropy of information.

The entropy of information

Entropy measures the disorder in a system. The heat death of the universe, isn't, as I thought, a conflagration. It's a room temp cosmos. With all the heat evenly distributed, no work can be achieved. The story's over. Bringing about a text's "heat death" is the writer's job. You take initial conditions which have some imbalance in them, some potential for energy to transfer. (Alternatively, you destabilize a balanced system.) Then you run the simulation, watching how the energy flows and diffuses within it, until there's nothing left to be changed or done or learned, and They All Live Happily Ever After. A physicist would call the fairy tale ending a state of thermodynamic equilibrium, or one of maximal entropy. "A system that is in thermodynamic equilibrium experiences no changes when it is isolated from its surroundings."

Entropy is polysemous, however, and information theory has its own definition: entropy is a measure of the uncertainty in a variable, and therefore a measure of its information content. So the more entropy a message contains, the more information it has. You can see how this clashes with the entropy of plot dynamics, in which the amount of information drains out of a system as the entropy increases.

Since I'm going to be talking about both types of entropy throughout this manual, I've chosen to refer to information entropy, or Shannon entropy, as "uncertainty."

Uncertainty is the resource contested in the metagame: the reader tries to gather as much predictive certainty as they can, and you're trying to keep it from them. If we reduce the metagame into a series of coin flips we hope the reader will not predict, what kind of coin do we want to be tossing? Say I present you with a series of coins, loaded to every ratio between 10:90 and 50:50 -- which would you pick?

Obviously you want the fair coin. A loaded coin only surprises over the short-run -- after awhile, the reader would catch on. Their best strategy would then be to guess the biased side every time, and they would make more correct predictions as a result.

That is our first principle: uncertainty is maximized when outcomes are equiprobable. Which is why the Immovable Object vs. the Unstoppable Force is such an exciting scenario. There's incredible suspense there. Something's gotta give!

But it should be easy to recall all kinds of fictional scenarios which were not fair coin flips. In fact, fiction loves the one in a million shot. Why?

Our second principle: the lower an outcome's odds, the more information it contains, and the more surprising it is. We love surprise. So while a loaded coin is not, on average, very uncertain, it will occasionally be quite surprising.

Uncertainty

Go back to our prediction game. Last time you chose the fair coin. Now I present you with a fair coin and a six-sided die. Which will be hardest to predict? The die, of course. It has more possible, equiprobable outcomes. In an ideal love triangle ABC, Anne should be equally drawn to Bill and Carl. Why not toss in Dave, Ethan, Frank, and Gary? Two reasons:

• the more outcomes, the more difficult it is to make them equiprobable
• diminishing returns.

Our ABC love triangle has one bit of information because it can be described through one binary question: will Anne pick Bill or Carl?

(If she favored Bill:Carl at 90:10, then it would only contain .46 bits of uncertainty. You can check out this graph from the Wikipedia article on entropy to see the other values.)

If we introduce Dave, and he's just as likely a suitor as Bill and Carl, the scenario now has 1.58 bits of uncertainty. That's good: more uncertainty gives a better metagame. But I believe that it's much harder to set up a 33-33-33 probability split than a 50-50. Say you botch it, and the odds are 10-10-80: the total uncertainty would be 0.92, less than the 50-50 toss-up. Even if you do get all 1.58 bits, there's still another tradeoff being made: each character gets fewer pages of development.

Three's not bad, but taken to an absurd extreme, you can see how uncertainty alone isn't enough: what if Anne was the focus of a love chiliagon (a 1000-sided polygon)? That's 9.9 bits of uncertainty! Wow, so many bits! Wait, who cares? Whomever she chooses, we would have no feelings about the guy: you can't characterize a thousand dudes. Furthermore, information entropy plays out on a logarithmic curve, which means the marginal utility tapers off quickly. The love chiliagon has 500x the suitors, and only 9.9x the uncertainty of the love triangle.

Our third principle: information entropy simply describes the information's quantities: it says nothing of its quality. Highly uncertain gibberish is of no interest to us. We want meaning, and uncertainty, and surprise. Luckily, these all correlate to some degree. But uncertainty and surprise can only multiply interest: they cannot manufacture it.

Surprise

Or can it? We are fascinated by dark horses and underdogs. Every two years, during the Olympics, we sit down and develop passionate, temporary rooting interests in sports we do not care about. All I need Peter Hammond to tell me is who is the favorite and who is the underdog. If the favorite dominates, we're satisfied. If the underdog guts out the win, we get something like the Miracle on Ice, or Rulon Gardner versus Alexander Karelin. The latter is particularly instructive.

In 2000, the Russian Alexander Karelin had not lost a Grego-Roman wrestling match in thirteen years, and hadn't surrendered a point in six. His nickname was "The Experiment" and he trained by mushing through snowdrifts. His body was country strong, and his face seemed capable of either an iron curtain scowl or an insane grimace, with no in between. Pointed head, shaved for what I imagine were pragmatic reasons -- no purchase. In short, he looked every bit the Siberian juggernaut.

Rulon Gardner was born in America's Siberia: Wyoming. He is four years younger than Karelin, and in 2000, had no international medals to his credit. (Karelin had 24 international golds.) More importantly, Gardner was doughy. Big love handles, soft chest, no definition. Even setting aside the resumes, a single frame of the two men standing next to each other fixed the odds for any spectator: Gardner was a +1000 shot. Then he won.

It was a great Olympic moment, which meant it was kind of flukey. But how flukey was it? Judging from a photo, sure, Gardner looked overmatched. But in motion, it's easy to see that Gardner was also country strong, which doesn't always look very strong. And he was fast with his hands, athletic. After the 1-0 upset, he turned a credible cartwheel.

A fourth principle: the "real" odds of an event do not exist in the absolutely deterministic world of a text. The odds are interpreted by subjective observers, and so it is your reader that determines how surprised they will be, how uncertain any outcome is.

Your job is to help them set the proper line. Here are some ways you can do this.

Fixing the Match

If your only goal as an author would be to ensure that your reader never predicted the action, that would be easy -- lie. Better yet, tell them nothing; string together random occurrences and call it a novel. In one scene, a piano squashes a character. In another, a character spontaneously combusts. A beautiful girl hanging laundry in her back yard floats off into the sky. Et cetera.

In that text, the reader would never guess right. But they would never guess at all. They would have set the book down. To play the metagame, the writer has to provide (or give the illusion of providing) as much information as possible while still surprising the reader. This means delivering to the reader a reasonably accurate picture of a scenario's possible outcomes and their probabilities.

Which explains the common wisdom about motive. In an agon, the reader has to know what the character wants and why, so that they might anticipate the narrative's future. (I think it's crucial that the reader always has a guess, however wrong, of what will happen in an agonistic story.) Luckily, we can add more motives as chaff. If we have a character distinguished only by her professional ambitions, then her response to a plum offer of a gig overseas is assured. But if she is also a family woman, then her goals -- both authentic -- now flummox the reader a bit. Here's a crucial lesson: more data does not always result in more clarity. So by flouting the Gricean maxim of quantity, we can satisfy our lightly-opposed goals of forthrightness and uncertainty. The other maxims can be used as well.

The maxim of quality: "Do not say what you believe to be false. Do not say that for which you lack adequate evidence." A writer should respect this; but feel free to coax the reader into drawing false conclusions from inadequate evidence. This can be achieved only to the degree that you limit interiority. Once you let the reader into the character's head, good luck finding anything to talk about which doesn't tip off the character's motives. But if the reader has no access to the character's head, they must draw conclusions based on behavior. No reader considers this unfair; this is how we must interact with other people all the time. In fact, we all believe we're pretty good at it! And so we can again satisfy our goals of forthrightness and uncertainty, this time by providing all the behavioral data but none of the impulse data.

The maxim of relevance: "Be relevant." The red herring -- it's a classic for good reason. It works because the writer sets the terms of the discourse, and so the reader can't be sure what's relevant and what's not. If I ask you where 4th street is, I know you're being irrelevant when you tell me a story of your childhood. But who knows what the writer is driving at on page one? The Silence of the Lambs fake-out is one of the all-time great bits of irrelevance. Gabriel Moura has a nice video on YouTube called Example of Parallel Editing in "The Silence of the Lambs". To summarize: the director shows us a SWAT team poised before a door, ram at the ready. He then shows us a serial murderer inside a house. We assume that the two must be connected, but in fact the SWAT team is not at the murder's door at all, and it's the protagonist who has stumbled into the lion's den. Again, we have an aboveboard amount of data, but it's deployed trickily. Our assumption is that contiguous shots belong to the same space and time. The lesson here: signal can be drowned out by noise just as well it can be muffled by silence. Further: the reader tends to assume that everything is signal.

The risk with these red herrings is when they're obviously such. The sooner the reader figures out the red herring, the longer they sit tapping their foot, waiting you to reveal it: not good. Therefore, a red herring should serve one other purpose besides plot obfuscation. If a reader figures it out early, at least they can think, "Well, this is all irrelevant to the main plot, but at least it's fun." (Sorry that my recommendation is to write better. I try not to do that.)

The maxim of manner: "Be perspicuous. Avoid obscurity of expression. Avoid ambiguity." Perspicuous is a great word. It has to do with translucence, clarity. I think most writers strive for it. We want sentences that are like clean windows: transparent, admitting light, providing a good view of the world. But when it comes to plot, the hell with that.

You can see this maxim flouted by that proud lineage of prophets, soothsayers, mediums, and clairvoyants that populate our fiction. Do you know why a real-world fortune teller takes pains to be cryptic? Because they can't see the future, and the Forer/Barnum effect means they're better off being vague. But fictive fortune tellers actually can see the future -- if the writer deigns to allow it. Still they are cryptic.

That's because information without context is noise dressed as signal, and this noise can be used. Breaking Bad has given us a few glimpses of the future, in the form of flash forwards, and these give Gilligan a powerful tool for priming our expectations, while simultaneously not revealing anything significant. At the start of Breaking Bad's second season, we see a pink stuffed bear floating in a pool. One eyeball is missing; one side is charred. What's all that mean? Who can say, it's ambiguous, and it's intriguing. What it really means is that the reader will be watching like a hawk for a pink stuffed bear or anything flammable.

Other Techniques

Yomi

A game designer named David Sirlin has an idea called yomi. It means "reading" in Japanese, and it boils down to something like this: they don't know that we know they know we know. When you're playing a game with someone, how does your knowledge of their strategy affect your own? How does their knowledge of your knowledge of their strategy affect both of your strategies? These chains of moves and countermoves can get paralytically complex, but for writers, we only need two moves to create suspense: one obvious move and its reverse. Let's take for an example the "I'm one day from retirement" guy. In the real world, I imagine that cops one day from retirement do not die at unusual rates. In fictional worlds, that is quite literally a death sentence. Readers have seen enough movies to know something terrible is about to happen. Which can be frustrating: you want some set up for your doomed cop, but you don't want to telegraph it. On the other hand, it's great when you can exploit readers' training. If I know that my reader knows "one day from retirement" leads to dead cops, I can fake her out by having a cop say that. Then, the reader will think it's too obvious to kill the guy, and the cop is safe, or that it's so obvious to kill him that it's not obvious, and he's doomed.

The brilliant thing about these prediction games is that you can get the reader fixated on the wrong thing while you steal his wallet. Maybe while they're handicapping the cop's death, they're missing the clues that point to another character's imminent demise.

Paradoxically, the less savvy the reader, the harder they are to fake out, simply because they're insensitive to some of your misdirection. As Sirlin points out, though, once you realize you're dealing with a rookie, you can take that into account. Imagine how easy it would be to write plot twists for an audience of little kids.

Thinking, Fast and Slow

The metagame requires analysis, and the easiest way to win the metagame is to short circuit the reader's powers of analysis. Daniel Kahneman wrote a good book called Thinking, Fast and Slow, which describes two systems of cognition. System 1 is fast, sloppy, effortless, and emotional. System 2 is slow, precise, laborious, and rational. The more the reader is in System 1, the worse their predictions will be.

It sounds a little sinister to talk about controlling your readers' headspace, but that's what any text is doing, and besides, this is what the reader wants. They want to get wrapped up and emotionally invested; the sloppy probabilistic thinking attendant to that is just a bonus for us.

Engaging System 1 is hard, though, and often we make it harder on ourselves by half-assing it. Personally, I've never been too eager to create a character that's very easy to root for or against -- it seems unrealistic to me. But that's just the most expedient way to hook the readers emotionally. Great characters can inspire emotional investment and be thoroughly human. Of course if you can pull off that kind of great character, you don't suffer from any of the problems I'm exploring in this piece.

While I recommend the whole book, the Wikipedia page for heuristics gives us enough to chew on. A heuristic is a cognitive shortcut, the opposite of an algorithm. System 1 relies on these. Here are a few we can use:

• Anchoring effect: we make judgements relative to the first piece of data. This is why the salesman starts high while haggling. In a story, you can have the characters anchor the reader's expectations by offering their own opinions of what will happen.

• Availability: our certainty in a belief depends on how easily we can think up the evidence. If an antihero's been on a villainous streak, we won't predict decency, even if he's been on the whole more good than bad. Same thing if he's done one spectacularly bad thing. The magnitude of the bad deed overwhelms the many smaller good deeds. Also the reason why Chekhov's gun works. Since no fanfare accompanies its appearance in the story, it is difficult to remember it.

• Representativeness: "The representativeness heuristic is used when making judgments about the probability of an event under uncertainty." (Sound familiar?) With this heuristic, the closer the known data fits a particular stereotype, the more likely we are to fill in the unknown data with information from that stereotype, even if that violates the base rate. This heuristic is why Bayes Theorem can be so hard to wrap your head around. A good example from Wikipedia:

John is a man wearing outstanding goth inspired clothing with long black hair who listens to death metal. How likely is it that he is a Christian and how likely is it that he is a Satanist?

If people were asked this question, they would likely underestimate the probability of him being a Christian, and overestimate the probability of him being a Satanist. This is because the base rate of being a Christian (there are about 2 billion in the world) is vastly higher than that of being a Satanist (estimated to be in the thousands).

How do we apply that to stories? We can use it to obfuscate character's true natures. Mad Men's pilot is a textbook case: Don acts like he's unmarried, but in that era, what percentage of men his age (and with his income) were bachelors? Even Sal, who's gay, is married. So the base rate for bachelors must have been pretty low.

And still my mind was blown when Don comes home to a suburban house at the end of the day and kisses his blonde wife's forehead.

Patterns useful for predictability

Patterns make probability. Many of the odds-setting techniques I've mentioned revolve around credibility. I, the writer, tell you something and hope you'll buy it. With patterns, I can just show you events, and allow you to extrapolate the base rate from your observations.

This is an unusual support of "show, don't tell." If you have an underdog character, it's conventional to show him getting kicked around. Why? The cult of conflict provides the typical rationale: conflict is always more interesting than non-conflict. Is it? Watching a couple throw plates isn't innately superior to watching them fall in love.

I believe that the value in a character's tribulations has a lot to do with the metagame. The heartbreak and failure set long odds on the later triumph, providing the illusion of difficulty.

Which is ridiculous, of course, when both reader and writer know the ultimate outcome of a story, and they collude in pretending they don't. Can we have an agon, can we have a metagame, if the final outcome is a fait accompli? Yes -- that's the magic of competition. I'll watch basketball games whose outcomes I already know, provided the action is that good.

Readerly Countermeasures

Despite all these tricks at our disposal, the metagame is somehow still tough to win. How could we fail to blindside these suckers?

Alas, the reader has outside information. They know how the real world works: what is and isn't possible. They started consuming fiction not long after they figured out the toilet. They know the conventions of drama, as well as the conventions of whatever mode or genre you are working in. They know how long the text is! That last piece of intelligence devastates our subterfuge. I don't care how suspicious a suspect you've created -- as long as the reader's got two inches of paper between finger and thumb, they won't bite. Same reason the antagonist always escapes the first encounter. There's too much time to go for the main plot to resolve now. (Potential workaround: maybe the "main plot" is actually a diversion.)

Though we can't stop our readers from collecting this outside information, we can still take advantage of it. Let's see how some other writers have done just that.

Order of the Stick and the Rules of Magic

Order of the Stick is a stick figure webcomic that enjoys an baffling popularity. People love it, and it's not clear why at first glance. The art is fine, as far as stick figures go, but the resolution is poor, the compositions are cramped, there's tons of text... the visuals aren't the selling point, clearly. One might next assume there is something wonderful about the characters, but they all kind of sound alike, beyond slight variations in accents and diction. The jokes aren't great, tending to be fourth-wall stuff or pop-culture gags.

But Burlew is a good plotter. He's meticulous about set-ups, and tantalizes his readers with long and short-term mysteries. (The forums are full of prediction threads, some having to do with mysteries that are a decade old.) The metafiction's true value isn't to enable cheap jokes, but to bring the metagame into the open. There's another unusual advantage, one that Burlew probably didn't anticipate at the outset: by basing his story in the universe of D&D, he and his readers share massive tacit knowledge. This encourages the readers to predict outcomes as if they were the characters. I've read long forum posts in which readers analyze the heroes' predicament from the perspective of the RPG min-maxer. As an example:

Off hand I know of no fighter feats to help Roy here. Certain Pathfinder barbarians could stand for a long time inside something like this though [...]

How great is that? What seems to be a disadvantage of writing a story from within an RPG is actually a feature. The limited range of possible actions means that the expert reader can do a full analysis of any scenario. And this analysis has nothing to do with drama, really -- this is treated as a physics problems. These laws being immutable, what might this character do here? The simplicity of the D&D model gives readers the chance to be minor gods peeking over the shoulder of Burlew as he builds predicaments.

In a different kind of story, where anything can happen, they won't bother to do this -- there are too many possibilities.

Breaking the rules

The content of fantasy does not excite me: elves, swords, and spells get at nothing that cannot be reached by some other signifier. It is fantasy's access to the impossible that excites me. Magic can bring readers to the spirit's hinterland; there, all this nonsense about agons is suspended. Faced with the truly unknown, prediction stops: we simply pay attention. I think that is refreshing, as a reader. As a writer, I milk it for all it's worth.

Because the metagame never stops, and you can break rules to find great moments. Remember when Indiana Jones shoots the scimitar-wielder? Why is that so memorable, so unexpected and delightful? The suspension of disbelief demands a massive, unacknowledged act of collaboration on the reader's part. Since fiction often seeks to mimic reality, we sometimes forget how divergent they are. So when the authentic erupts from all the other ersatz authenticity, it's bracing. Were any of us in that bazaar, watching Indy face down the guy with the scimitar, none of us would be surprised at the outcome. (Why? Other fiction. The Untouchables taught us "you don't bring a knife to a gun fight.") But typically, whenever the hero is challenged to a duel, he must respond honorably, partly because it's convention, and partly because it's more exciting to see a slobberknocker than an anticlimax.

The value of a broken convention doesn't mean you should ignore the laws of drama. They are conventional for good reason. All that's required is a single, shocking subversion. Once you establish that you are a writer who is willing to forgo a spectacular scene for the sake of authenticity, the reader has a much more difficult time predicting your moves. You are like a dog that once bit someone. A Song of Ice and Fire breaks a major law of drama in the first book; as a result, you're always expecting some trickery, even though George RR Martin is generally straightforward in his outcomes... well, at least until your guard drops again.

Conclusion

Whew. We made it to the end. A writer who has mastered the agon and the metagame will get his or her readers to the end, too: all these techniques I've been musing about add up to a page-turner. But a page-turner isn't always a good book, and we should be suspicious of anyone who thinks they are equivalent. None of a text's artistry should be sacrificed for the sake of the text's addictiveness or suspense. Satisfaction over surprise, always.