Over the last few years the IF community has become more systematic about how we talk about structure in branching choice-based narrative. Sam Kabo Ashwell’s Standard Patterns in Choice-based Games is a go-to article defining some useful terms; it pairs well with Choice of Games’ article on how to use stats to create long-term consequence without combinatorial explosion, and Jon Ingold’s talk on inkle processes at GDC 2015.
A lot of that conversation revolves around the shape of the whole plot, or at least whole chapters, though; so I wanted here to talk briefly about some structures that I find really useful at the smaller scale.
Confirmation-required Choice. One of the things Jon talks about sometimes is the use of text to let the player opt in to doing something profoundly stupid, through a series of escalating choices. Are you sure you want to do this? It looks like the monster is getting angrier. Are you still sure you want to attack? Yes? You notice that the monster’s bite is poisonous. Are you going to attack now?
Once the player has opted in multiple times, it’s really on their own head if they wind up in a situation with a combat roll that wipes them out. This expands what would otherwise be a binary decision into an experience with more tension; it also tends to work well in cases where there’s one dangerous-but-interesting option and one safe-but-bland option.
How to enhance with stats: count how long the player sticks to the risky path before giving up (if they give up). Use this later as a metric of their commitment to the dangerous cause, and/or their recklessness.
Track Switching Choice. A variant on the Confirmation-required choice is one where the player is allowed to change their mind in either direction for several beats. Later beats might introduce some potential drawbacks and warnings about whatever track the player is currently pursuing. Like the confirmation-required choice, this is a way to give some extra weight to a decision and emulate a situation where the protagonist might be genuinely conflicted about what to do next.
How to enhance with stats: track how often the player chose the outcome they ultimately landed on, vs. another option. Use this later as a metric of their commitment to a cause, or their willingness to change their mind about things.
Scored Choice. In the track switching choice, we hold the player to their final selection, whatever that might be. With the same basic structure, we could also score how many times the player chose one way vs. the other, and then use their top score to determine the outcome. The track switching choice often works well when there’s a single tough decision in the story; the scored choice is a good fit for a montage of related choices. For instance, if the player is choosing between a career-enhancing move or staying with a romantic partner, we might have a series of small decisions that test their commitment to one option or the other. (This strategy pretty much requires stats.)
Re-enterable Conversation Node. This one’s a classic standby in a lot of choice-based games; you’ll find versions of it in many Choice of Games works, in conversation nodes in Fallen London, and in Twine pieces where you can choose to talk about or explore sub-parts of the scene before moving on.
Even at its most basic, this approach can be a good way to liven up what would otherwise be a long, linear exposition dump. With a bit of artistry, it can be more interesting and exploratory.
How to enhance with stats: limit how many of the sub-nodes the player is allowed to visit, so they’re forced to see only a subset of the available nodes. Track which information they seemed to be interested in. (Texture explicitly provides for this by letting the author limit how many actions the player can take in a given scene before being forcibly moved on.)
Another option is to gate some of the topics so that they open up only when the player has learned preliminary information, or when the player/NPC relationship is in a particular state. Fallen London conversation nodes often use a “current topic” stat, so that some topical questions only become visible when the conversation has reached the right subject matter.
The next two examples work best in stats-based pieces like ChoiceScript or StoryNexus games, or Twine or Texture games that set variable during the course of play:
Chapter One Sorting Hat. A Sorting Hat (in Sam Kabo Ashwell’s definition) is a structure where the player makes a decision early on and then is sorted into one of several otherwise pretty linear paths. As a structure for a whole game, I don’t always find that very satisfactory, especially if it means that the piece becomes almost uninteractive after that first decision. (An exception: I thought Magical Makeover was delightful.)
But I do use the structure a fair amount when I’m writing the opening of a game I think the player might want to replay several times. Often, especially in stats-driven choice-based work, the player doesn’t start seeing significant variations in content until several chapters in, making the opening less fun to replay.
So for my in-progress Choice of Games work, I give the player the opportunity to explore a couple completely different character backstories in the early game. Each of those character backstories contains some internal choices that set stats and have consequences later in the game as well. But I’m hoping the narrative variation in the early game will help keep replayers engaged enough to see the more subtle stats-driven variations in the later game.
At the end of the chapter one sorting hat, you bottleneck as usual: this is really just a specific subcategory of branch-and-bottleneck that maximizes the immediate narrative impact of early-game choices.
The converse of this is the Endgame Time Cave. The Time Cave is the sprawly, every-branch-leads-to-new-outcomes structure, and, again, it’s not always the greatest choice for a full game. As a way to pay off consequences at the end of a story in a stats-based piece, though, it can work pretty well: looking back over the player’s accomplishments to date, then giving a narrative consequence to where each of their major stats ended up.
In the diagram here for simplicity I’ve labeled those outcomes as “good” and “bad”; in practice in a lot of stories this will be less simple than win/loss. You might have outcomes depending on particular relationship states, for example.