Jon Blow’s MIGS Keynote and Thematic Iteration

I didn’t get a chance to take in Jon Blow’s MIGS keynote from November until now. Well, I just listened to it.

First, you need to listen to this talk. I’ve heard (and said) a lot of ideas that orbit around much of what Blow discusses, but I’ve never heard these concepts crystallized so effectively. It’s worth listening to on it’s own, and you also won’t get this essay if you haven’t listened to it yet.

I’m going to offer a solution to Blow’s “hard problem” of creating an interactive system that coherently expresses meaning.

Blow says that games have more trouble than other media in conveying meaning because we have so many other things to worry about. For a coherent meaning to emerge, he says, the “dynamical meaning” expressed by the design and gameplay must dovetail with the meaning of the embedded story. This is difficult because we can’t predict a priori what the design is going to look like when the game is finished.

Furthermore, the player himself is an agent of chaos. He will destroy all of your carefully-crafted pacing and story beats. He will ignore important dialogue, or strip his clothes off before the funeral scene. How do we create an experience which will express our desired meaning with this madman running amok on stage?

The hole in Blow’s reasoning is that he seems to assume that we need to decide on the meaning of our games before we create them. He explicitly condemns the simple “message-centric” type of story, but he still thinks that authors need to have a priori control over the meaning of their creation.

They don’t. The first part of the solution is to let go of the idea that we can or must be able to make a game to convey an arbitrarily chosen meaning. As Blow correctly says, it’s nearly impossible. But this doesn’t mean we can’t add meaning to our game.

Other media have exactly the same sorts of problems in combining different elements which are pulling the meaning in different directions. A filmmaker may want to characterize someone with some dialogue, but what if the pacing demands an action scene? A novelist may want to characterize the doctor as the bad guy, but what if the plot requires him to save someone’s life? These creators don’t have to deal with a beast as unpredicable as an interactive game design mucking with their stories, but the problem is still there. And they’ve developed ways of solving it. So let’s learn from them.


Stephen King has sold 350 million novels. He practices freewriting – writing without planning and allowing the story to emerge during the writing process. The Dark Tower series was started when King simply sat down at a typewriter in 1970 and wrote a single sentence: “The man in black fled across the desert, and the gunslinger followed.” King wrote a seven-book megasaga from that one line, without planning any of it. When a dying woman asked him, before he had finished the saga, how the series was going to end, he couldn’t tell her. He didn’t know.

How does King create meaning, then? He iterates.

After he writes the initial draft, he goes back over it and tries to figure out what themes and motifs naturally emerge from the plot. He then tweaks the story to emphasize those concepts.

For example, King freewrote Carrie straight forward, without planning. After he finished, he read through his draft and noticed that blood appeared repeatedly in the story. He then went back through and emphasized the blood even more, adding it to other scenes, describing it more deeply, and imbuing it with meaning. Thus blood became a repeated motif, and concepts associated with blood became themes of the story. He didn’t know that blood would be important until after he had written the story.

We can do this in games. In fact, we already have. Blow mentions the Little Sister Adam harvesting mechanic from Bioshock in his talk. I work at 2K Boston and though I didn’t work on Bioshock itself, I’ve talked with the designers who were around when Bioshock was being created. For most of development, the save/harvest Little Sister mechanic did not exist. Originally, Jack could only save the little sisters, and there was no narrative reference to any kind of moral dilemma between surviving and saving the little girls. The save/harvest mechanic was only introduced late in development as a gameplay choice. It then became part of the story, and finally was threaded into the theme of the entire game.


Certainly, this could have been done better – Blow specifically criticizes the save/harvest mechanic because the story seems to contradict the design. I still think that the mechanic was a success, if only because so many people gushed so much about it. Imagine how much stronger the game could have been if the developers had chosen a theme that arose even more naturally from the design.

Iteration is one of our bread and butter tools, but we currently only iterate to maximize “fun”. There’s no reason, however, why we shouldn’t be willing to iterate on theme as well. Take a “theme pass” on the game in the last three months of development, just as Stephen King did with Carrie. Play the game, figure out what you think it means, emphasize those concepts in the art, the dialogue, and perhaps even the gameplay.

Imagine doing this for Left 4 Dead. Stop reading now. Think about Left 4 Dead as it exists. What is it about? What does the design want to say?

My first thought was to say that the game is about the inevitability of death. I wanted to make one of the characters a mortician and add a bunch of nihilisitc dialogue about the heat death of the universe. But this theme doesn’t fit. Death is not inevitable in Left 4 Dead. It’s not what the game wants to talk about.

Recognizing this, I started searching for another thematic seed. This is the beauty of choosing theme after finishing most of the design. If the design contradicts the theme, just change the theme!

Next, I chose to think that it is about how strangers come together (or are forced together) in a time of crisis. Now imagine we do one more development pass on the game to grow this thematic seed. The theme already exists naturally in the game. Now we’re making it much more important. Imagine dialogue revealing that Bill and Francis were enemies before the infection, but were forced into a friendship afterwards. Imagine that pills and ammo are given by wounded NPCs who donate their last supplies to try to help people they don’t even know. Imagine that the news chopper pilot and the boat captain are given dialogue to indicate that they are risking their own lives to come help the four player survivors, who are total strangers to them and to whom they owe no loyalty. Imagine that the chopper pilot dies if you don’t survive to the end of the No Mercy campaign. Imagine the guy in the church in Death Toll, instead of trying to kill you, tries to help you, but draws the zombies accidentally while doing so and dies because of it. Imagine making Bill a racist who doesn’t trust black people. Suddenly, the game isn’t just fun, it’s about something.

None of these changes would corrupt the design. Blow says, “Late gameplay changes are very expensive!” This is true. But you don’t need significant gameplay changes to emphasize a pre-existing thematic seed if you choose the seed well. It takes very little. My Left 4 Dead example doesn’t include any real gameplay changes at all – only some new animations, art, and dialogue.

We can do this for any game. They all have seeds of meaning that we could grow. We only need to listen to and emphasize what the design wants to say, instead of trying to force it to say something we decided on beforehand.

2 thoughts on “Jon Blow’s MIGS Keynote and Thematic Iteration

  1. JP

    This is the beauty of choosing theme after finishing most of the design. If the design contradicts the theme, just change the theme!

    While it’s good to have, if not employ, this type of flexibility, the classic Structure vs Dynamism argument from the other direction is that the more you treat theme as a brush-on veneer over the game design, the less chance that theme has of being deeply resonant. Bioshock did get a “thematic unity” pass late in development (as well as some simple plot hole fixing) but the later you do that the thinner the veneer will be, and the earlier you do it the greater the chance that you’ll have to rework everything, lather rinse repeat.

    Part of the reason I think most modern, AAA games are having such a hard time avoiding ludonarrative dissonance (Clint Hocking’s term) is because they have so many other demands placed on them: be a millions-selling blockbuster franchise entry, be a technological showcase, be a tight accessible piece of game design, tell an epic story – when you add “convey a theme that resonates at every level of the whole” to that, something’s gotta give.

    Look at Rohrer’s “Passage” on the other hand. Sure it’s an incredibly simple game with almost nothing to it but its theme, but that’s exactly what allows it to (in my opinion) convey its theme so elegantly and powerfully. It doesn’t have to sell millions or show off per-pixel lighting effects. Every single line of its code was built towards the game’s message.

    All that said, I agree that iterative exploration (a subject touched on in another Jon Blow talk) is the only way to go for developing a theme layer, and your suggestion, that dev teams use it the way they currently do for the “is this fun?” design process, is a great way to phrase it and might help more devs realize that “art game” isn’t some ridiculous rarified process but a simple difference of intent.

    No game design survives contact with reality unscathed, after all.

  2. Tynan Sylvester Post author

    I don’t think it’s necessarily true that doing a theme pass on a finished game design would make it a “thin veneer”. I think it more depends on how much you invest in the theme pass itself. If Valve did a year-long theme pass on Left 4 Dead, for example, I think they could make that theme pretty damn powerful – even if they didn’t think about it at all until the theme pass started.

    Of course I’m just supposing. I’ve never actually done a theme pass. I’ve tried several times to build games around a preconceived theme though, which never worked. So you could say I know what doesn’t work – now I’m trying to steal something that does from Stephen King.

    About the multitude of demands: it’s totally true. Novelists can focus on theme because their medium is so much more constrained than ours. When you can do essentially anything, it’s really hard to focus. I think a designer with a theme intent might do well to take a very very well-trodden genre (shooter, platformer) and restrict themselves to relatively generic mechanics within that genre. I’m sure there’s still tons of room to work within even the most overdone types of games.

    Thanks for the link, looks cool. And I’ve been thinking about the documentation masturbation (my term) problem recently, so it’s timed perfectly.

Comments are closed.