Balancing Narrative Design and Player Agency: Part Two

The Philosophy of the Branching Railroad

By: Heather Albano

“Let’s invent a world where the player gets to go through every emotional journey available.”                                                                                                                                                        – Clive Barker

“There’s no way…any company would have the resources to make a game that fully accounts for every player decision and branches the story accordingly while still hitting the same story beats.”                                                                                              – benjaebe,

It is not possible (today) to design to Clive Barker’s ideal. There really isn’t any way a company operating in the real world would have the resources to make a game that fully accounts for every player decision and branches accordingly.

But more importantly, that is not the no-pun-intended path to good narrative design. The more branches you are dealing with, the less of a chance you have of doing any of them well, and the less distinct they will be from each other. Designing your game as a branching railroad does not mean allowing the player to do whatever they like and branching the story accordingly. It means allowing the player to go through multiple (not infinite, but not singular) distinct and satisfying emotional journeys based on their choices. Under this philosophy, player agency dictates which one of multiple well-designed narrative arcs is experienced—i.e., whether the railroad takes them to Awesometown, Awesomeville, Lake Awesome, or Awesomeville Heights.

All Roads to Awesome-1

Let’s take a simple example first. If I were going to make the “star-crossed lovers” game with two possible endings, the Claudio and Hero one and the Romeo and Juliet one, the first thing I would do is look for the choice that caused the difference. I think there’s a good argument that the first story plays out happily and the second plays out sadly because Friar Francis is a whole lot more careful than Friar Lawrence when setting up his conspiracy. Admittedly, neither friar is a main character, which weakens this argument a little, but both are absolutely active characters, actors who affect the stories in which they dwell—what Shakespeare might have called players in their dramas. The friar-player takes an action (makes a choice) that determines whether the narrative arc is a tragedy or a comedy.

It happens well before the end scene, but the story doesn’t branch there. The game wouldn’t need to branch there either, only set a variable that remembers what the friar did. So you’d design it this way: At the point where the friar sets up the conspiracy, the game stores a variable recording the level of care. The rest of the game proceeds linearly. And then, at the very end, the variable is called, leading to one of two end scenes—one that makes the story a comedy, one that makes the story a tragedy.

The game designers would not need to craft two completely separate branches, only a single additional scene. All the other scenes—falling in love, beset by external forces, separated by those forces, clever plot, everyone thinks the girl is dead, all seems to be lost—are exactly the same. The despair on Claudio’s face is the same as the despair on Romeo’s. But in one arc, the moment of “all seems to be lost” is the suspense before the happy ending; in the other, it is a beat in the tragic downward curve. The two different endings change the player’s perception of all that came before.

Now let’s take a more complicated example. (Serious spoilers for Choice of Broadsides follow, but the game’s been out since 2010, so…)

In Choice of Broadsides, a game I co-wrote with Adam Strong-Morse and Dan Fabulich, you play a naval officer of the Napoleonic era, in a secondary world, fighting for “Albion” against the “Gauls.” A secondary world Horatio Hornblower, essentially. Very early in the game, you are involved in taking an enemy prize ship, whereupon you meet Lt. Villeneuve. In the middle of the game, you meet Villeneuve again, during a temporary peace between your countries. At the end of the game, you face Villeneuve in battle. These points are hardcoded.

In the first encounter, you can treat your prisoner as well as honor demands; you can treat him even better than honor demands, so very well that your crew objects; or with polite chilly distance; or very badly and dishonorably indeed. Depending on what you choose to do, a variable is set.

In the second encounter, that variable affects what choices are open to you. Your actions here can take the beginnings of a friendship or enmity and dial them back to chilly polite formality, or your actions can deepen your mutual friendship or mutual hatred. At one extreme end of the scale, encounter 2 ends with you and Villeneuve fighting a duel at dawn; at the other, you and  Villeneuve end up banding together to fight off footpads. Depending on how you set your gender and sexuality variables up front, the footpad fight can even end with the option of a love affair.

So depending on what you do in encounter 1 and encounter 2, when you meet Villeneuve in battle in the climactic scene, you are facing either your star-crossed lover, the person who could have been your dearest friend if not for the unfortunate fact of war separating your countries, your oldest rival (in an intellectual sense, mostly functioning as “the face of the opposition”), or your bitterest enemy with whom you have many personal scores to settle. All four of these are recognizable narrative arcs that feel satisfying; all four of them feel distinctly different from each other; and all four are driven 100% by upstream player agency.

The climactic battle scene is the same scene for all four arcs. The player has the same options within that scene no matter how they got there, they have to perform the same actions to win the encounter, and the same stat checks determine whether they won or lost. But it looks completely different to the player who comes at it from the “star-crossed lover” path than it does to the player who comes at it from the “I’ll make you pay if it’s the last thing I do” path.


As Daniel Greenberg phrased it at GDC 2014, “The game world does not exist on the screen. It exists between the ears of the player.” The player’s perception of any given scene is influenced by their previous experiences; their recollection of earlier scenes is influenced by the one they’re in. This shifting perspective is a tool that the designer can consciously use.

If you plan it out ahead of time, you can make the same scene function as a certain beat in one arc and a different kind of beat in another. The “oh God, she’s dead” moment functions as “the darkest moment is just before the dawn” in Much Ado and a downward step in the unrelenting tragic arc of Romeo and Juliet. The first and third encounters of the Villeneuve arc are the same no matter what you do in the second, but the variable nature of the second changes the entire story.

If you default to designing for reuse, you have more flexibility to briefly branch in the places it matters, with an only incremental—instead of directly proportional—additional cost of production resources. However, this does mean you have to architect the whole thing up front—the entire game and all arcs through it—and that task carries its own complications.

We will talk about the nuts and bolts in the third and final installment of this series.

Heather Albano is a storyteller. Sometimes she writes traditional fiction and sometimes she makes games, and she finds the line between the two growing fuzzier all the time. Her most recent game, A Study in Steampunk: Choice by Gaslight, incorporates all the principles talked about in these blog posts, and is available here.

If you want to know what she’s doing next, or just follow the random thoughts that wander through her head, check out

Leave a Reply to Anonymous Cancel reply