Game Theory and When to Kill Off a Character

Posted on by Lopt and tagged , , , , , , , . Bookmark the permalink.


Just to make things interesting, I’ve decided to number the points according to the chorus of a catchy song.


Ever since people have been able to create works of fiction, they have sought to relate it to the external world in some way. Perhaps the story was actually about the real world, perhaps it was only loosely related, pausing to grab some basic physics or some geography before careening off into fantasy or metaphor. Usually the author can’t resist but to include Life, in some form or another, and all the things associated with it. One obvious such thing is Death: whether it’s to make some sort of philosophical point, move the plot along, or increase the reader’s personal involvement because they identify with the character(s).

The problem is that Death is often used so shabbily by the creators of fiction. It happens for reasons of bad writing: Death occurs in ridiculous or clichéd ways[1], or it seems to only affect the secondary supporting cast, if indeed anyone dies at all[2]. Now I can’t make you a better writer with regards to the first part, but I can tell you (with math!) how to go about fixing the second.


The idea is to use something called “game theory”, which was invented by Russell Crowe during the filming of A Beautiful Mind, which was a movie about how the government will implant false memories in the minds of academics in order to push them towards schizophrenia[3]. Writing has some similarities to games of deception, because the writer has to obscure how the plot is going to evolve so that it’s interesting enough to read. As a model, we imagine a game kind of like poker, with two players, A and B, and three ranks of cards labeled 1, 2, and 3 (with 3 beating 2 and 2 beating 1). The game works with the following rules: A can bet or check (not bet anything) and B can either call or fold. In this example, suppose that B always gets card 2, and A gets either 1 or 3 with equal probability.[4]

So the naive strategy for A would be to always bet when he’s got a 3 and to check with a 1, because he doesn’t lose anything due to lousy cards and stands to gain if B calls him when he’s got a 3. That strategy would be ineffective, however, because B would fold when A bets and thereby ensure that A could never achieve a profit.

A better strategy for A would be to bet with 1 a certain amount of the time, but always bet when A has a 3. This way B would be tricked into both folding his 2 to A’s 1 (a bluff) and calling A’s 3 with his 2 (trapped).

I’m at a place called vertigo

What does this have to do with fiction? Well, it should be pretty obvious: if you never inflict meaningful suffering/consequences/random calamities on your characters sufficiently frequently, it will mean nothing when they are in harms way. Similarly, if they never fail to achieve the objectives or goals that the story sets up (beat the bad guy, get the girl, etc.), it will mean nothing when then they do in fact succeed. You should have bad things happening more or less inversely proportional to how bad they are (within reason). Ominous threats and undertones of danger are only effective if, sometimes, they’re followed through on. This is what makes Tom Clancy a shitty writer and Ian Fleming a great one, even if they’re dealing with similar material. Once you take away personal involvement with the characters, you’re left with your prose or the engaging nature of your plot to keep people interested, which is a dicey gambit indeed.

So please, stop killing off the guy with the red shirt, make the stakes real. Not just because some random guy on the Internet suggests it, but because mathematics demands it. After all, you can argue with me, but you can’t argue with Russell Crowe[5]

—- Footnotes —-
1. I watched an episode of Criminal Minds a couple weeks back where a guy who’d been staging auto accidents to make his murders look less suspicious was chased onto a highway and subsequently dispatched by a truck. It practically made its own post-mortem one liner and thereby ruined any dramatic release that might’ve been possible if he was apprehended or at least eliminated in a less obvious manner.
2. Probably not a good thing that we are as a collective audience people who find the concept of a major character (and not even the main character) dying in the course of their story to be distasteful and alien.
3. Those interested should rent A Beautiful Mind‘s sequels, Men In Black and Men in Black II for further development of that idea.
4. Radically simplified for people who don’t give a shit about math.
5. He’ll throw a phone.

Related posts:

  1. Nobel Prize winning author thinks women can’t write.
  2. The end of original ideas, until the new one
  3. Dr. House Tells You What Life is Worth
  4. The Fast and the Fatherless

7 Responses to Game Theory and When to Kill Off a Character

  1. Pastabagel says:

    What you’ve described is basically why TV sucks, and why most genre films suck. But you’ve also described the problem, which is the audience. Audiences, as much as they claim the contrary, actually like the formula. Sticking to the formulas mean not having to pay attention to the details.

    In the finale of the Sopranos, Tony dies. He dies right there, in the last frame. He dies in exactly the way that the show told us he would die: Bobby Bakala said something to the effect of “You won’t even hear the shot when it comes,” And when Phil Leotardo was shot in the passenger seat of his car, his wife driving didn’t react to the shot or his death spasm for a full two seconds. It was conspicuous. You were supposed to take note of these little details at the time and remember them (and many others). Tony dies at the end of the Sopranos. It’s justice.

    And yet, fans of the show reacted viscerally to this suggestion, concocting other scenarios to explain the cut to black (not a fade, a cut). They didn’t want him to die. But of course Tony should dies. He’s a horrible monster of a person. He has no redeeming qualities at all. Why would you want the monster to live? Because the fans who wanted him to live wanted to be like Tony, in some way or another. His death became for them a moral judgment. On them.

    Likewise, people hated No Country For Old Men, because Josh Brolin unceremoniously dies two-thirds through the film. It didn’t match the formula. He’s the main character, the good guy. You don’t kill the good guy and then have 30 minutes more movie. And the rest didn’t make any sense to them.

    But Josh Brolin wasn’t the good guy. He’s a bad guy also. That was the whole point of the story. If you paid attention, you would have noticed how the filmmakers introduced the characters in an almost parallel fashion. The first line that Brolin speaks is the same as the line that Javier Bardem speaks at the end of the scene that comes immediately before: “Hold still.” And Bardem kills people with a gun used to slaughter animals, and we are introduced to Brolin while he’s hunting animals. They are the same, just to a different degree.

    I like the game theory approach to this, too.

    • eqv says:

      I’m with you on the Sopranos but not No Country. Brolin (I think the name of the character is Llewelyn) isn’t a bad or a good guy, just an ordinary guy. Until he decides to takethe money, until he decides to let greed etc dictate what he does, until the money becomes more important than the safety of his wife and his mother. It’s inevitable that punishment is coming his way, because he broke his moral code. I think this is more obvious in the novel than in the movie.

      OP: I know exactly what you mean. No investment in the character means there’s no real point in reading (or watching). When James Bond gets the shit beaten out of him in Casino Royale (I haven’t seen the movie) it’s gruelling as fuck. The stakes are real.

      Also, your subheadings/footnotes were hilarious.

    • Comus says:

      I prefer a similar thing in Coen brothers’ Burn After Reading, when all of a sudden [SPOILER ALERT] in the middle of the film Brad Pitts aerobics instructor character is shot in the head and killed in the closet. Which of course is hilarious not because of his character but because he’s actually all along “Brad Pitt”, to whom nothing bad can happen. I haven’t heard of anyone say that “Oh my god, they killed Chad!”, but always “Oh my god, they killed Brad Pitt!”.

      This game theory set up with loads of anxiety has it’s most horrible and best example in Caspar Noés Irréversible. It breaks all the rules and does not spare the viewer. I highly recommend looking it up (if you don’t mind anxiety provocament^2)

  2. sdenheyer says:

    Case in point: the second to last episode of Game of Thrones. (When reading the books, I became far more heavily invested in the story after that event).

    • rapscallione says:

      I was just thinking about that. I haven’t read the books, just watched the series, but I almost had the opposite reaction: People had been dying the whole time, and while they weren’t necessarily main characters, they weren’t red shirts either. After… what happened happened (I don’t want to spoil it for anyone), I just assumed that everyone, at some point in the story, was going to die. It kind of removes tension for me, and makes it a bit harder to become emotionally invested in somebody, because there’s a huge probability they’re going to die.

      I just think Martin took it too far to the extreme of character killing. I’m all for it, I’m all for realistic endings, but too much of a good thing is… well, you know.

  3. vprime says:

    This is why I have always counciled my creative writing students not to set their stories in dreams. There are no real stakes in a dream, so the entire story is pointless. Similarly, though, I tell my students not to just kill off the main character because they can’t think of any way to end the story.

  4. jaime omar yassin says:

    What’s interesting is the social pecking order that fiction has to observe when visiting violence on the characters. There must be meaning attached to the suffering of minor characters, for the major character to advance in the plot. In a conventional framework, those who suffer must be the property of the main character in the mainstream terms–women and children. And those who are not property, must be characters made to be uninteresting, otherwise, they would be the main character. Those are usually relegated to oblivion by class and race. In some way, the way we write fiction duplicates the way we shape our societies.