Standard Patterns in Choice-Based Games

When I was analysing the structures of CYOA works a few years back, I began to recognise some strong recurring design patterns. I came up with some home-brewed terminology, but didn’t ever lay it out in a nice clear way. This is a non-exhaustive look at some of the more common approaches, somewhat-updated (a lot has changed since then).

I should stress that these aren’t discrete categories: while a lot of works will fall very straightforwardly into a single pattern, many will involve elements of multiple patterns. (And yes, I’m aware that you can often simulate one using the mechanics of another. That’s mostly beside the point.) Also, the example diagrams I’m using are smaller and simpler than would be likely in actual works.

Time Cave. A heavily-branching sequence. All choices are of roughly equal significance; there is little or no re-merging, and therefore no need for state-tracking. There are many, many endings.


Effects: The time cave is the oldest and most obvious CYOA structure. It is often good for narratives about freedom and open possibility, adventures that could go anywhere, flights of fancy. Time caves tend to have relatively short playthroughs, but strongly encourage replay: they are broad rather than long. Even with multiple playthroughs, most players will probably miss a good deal of the content.

The time cave’s structure is both organised by chronological progression and detached from it. It’s ungrounded by regularity: possibility is so open that it often becomes fantastic or surreal, with different branches occupying wholly different realities. The player has velocity but little grasp, vast freedom but little ability to comprehend it.

Examples: Edward Packard’s earlier work (The Cave of Time, Sugarcane Island), Emily Short’s A dark and stormy entry; Pretty Little Mistakes.

Gauntlet. Long rather than broad, gauntlets have a relatively linear central thread, pruned by branches which end in death, backtracking, or quick rejoining. The Gauntlet generally tells one anointed story, which can be adorned with optional content or prematurely ended with failure; if there are multiple endings, they’re likely to derive from a Final Choice. Gauntlets rarely rely on state to any great extent (if they do, they are likely to evolve into a branch-and-bottleneck structure.)


Effects: The player is likely to realise that they are on a constrained path, but the presentation of side-branches matters a great deal – do they mean death? incorrect answers? travel back in time? blocked paths, footnotes,  or scenic details? Most often, the gauntlet creates an atmosphere of a hazardous, difficult or constrained world. Sometimes this can be punishing or depressing; sometimes it can be darkly comic; sometimes it’s a sign that you’re in a work heavily dependent on reflective or rhetorical choice. Perhaps the easiest structure to author, gauntlets can be conceived of in similar terms to linear stories, and ensure that most players will see most of the important content.

There are two major varieties of gauntlet: deadly and friendly. Deadly gauntlets mostly prune the tree with failure; friendly ones mostly do so with short-range rejoining, and look a bit more like simple branch-and-bottleneck structures. Friendly gauntlets have been vastly more common in recent years, making up a high proportion of Twine works.

Examples: Zork: The Forces of Krill. Our Boys In Uniform.

Branch and Bottleneck.  The game branches, but the branches regularly rejoin, usually around events that are common to all versions of the story. To avoid obliterating the effect of past choices, branch-and-bottleneck structures almost always rely on heavy use of state-tracking (if a game doesn’t do this, chances are you are dealing with a gauntlet).

Somewhat rarely, the bottlenecks may be invisible – the plot branches and never reaches an explicit rejoining node, but the choices at the end of each branch are the same or similar, creating an exquisite-corpse effect.


Effects: Branch-and-bottleneck games tend to be heavily governed by the passage of time, while still allowing the player fairly strong grasp. The branch-and-bottleneck structure is most often used to reflect the growth of the player-character: it allows the player to construct a somewhat-distinctive story and/or personality, while still allowing for a manageable plot. There’s a tendency – not a necessary one, by any means – for playthroughs to be very similar in the early game, then diverge as the effects of earlier choices accumulate. In order for the approach to work, it has to be used in a fairly large piece; you need time to accumulate change before producing results that reflect it.

Examples: This is pretty much how Long Live the Queen works, and is the guiding principle of Choice of Games (Dan Fabulich uses the term delayed branching). It’s also a common plot structure in non-IF games that allow significant plot choices.

The quest structure forms distinct branches, though they tend to rejoin to reach a relatively small number of winning endings (often only one). The elements of these branches have a modular structure: small, tightly-grouped clusters of nodes allowing many ways to approach a single situation, with lots of interconnection within each cluster and relatively little outside it. Re-merging is fairly common; backtracking rather less so. Quests generally involve some level of state-tracking, and do poorly when they don’t. The minimal size for a quest is relatively large, and this category includes some of the largest CYOAs.


Effects: This mode is well-suited for journeys of exploration, focused on setting; the quest’s structure tends to be organised by geography rather than time. Indeed, most works of this kind involve a journey with a specific purpose in mind. Quests work well for grounded, consistent worlds, but within that context the player-character’s situation is constantly changing. The narrative tends to be fragmentary or episodic, like old-school D&D encounters: little chunks of story which might not have any great significance for the big picture.

Examples: The Fighting Fantasy books and their descendants (Lone Wolf80 Days).

Open Map. Even though quests are structured by geography, time still plays an important part: there’s a built-in direction of travel. But take a CYOA structure, make travel between the major nodes reversible, and you have a static geography, a world in which the player can toodle about indefinitely. Often this is a literal geography and relies on extensive state-tracking both explicit and secret for narrative progress. But it’s not an uncommon mode for things with assumptions grounded in the hypertext-novel idiom – static but non-linear works like Le Reprobateur.


Effects: This is often used as an imitation of the default style of parser IF, although some may be parallel derivation from the former’s D&D roots. As with classic map-based parser games, the narrative tends to become slower-paced and less directed; the player has more leisure to explore and grasp the world, but spends less of their time advancing the story.

Examples: DuelmasterChemistry and Physics.

Sorting Hat. The early game branches heavily and rejoins heavily (branch-and-bottleneck is a likely model here), ultimately determining which major branch the player gets assigned to. These major branches are typically quite linear – sometimes they look like gauntlets, but they might be choiceless straight-shots. Sorting Hats almost always rely substantially on state-tracking in the early game, and often bottleneck at the decision point.

Effects: The Sorting Hat is a compromise between the breadth of more open formats and the depth of linear ones. Sometimes the nature of the various branches is signaled to the player; this is kind of important, in fact, because the player is pretty likely to notice the linearity of the second half and might assume that all of their choices will ultimately get funneled into that particular thread. The player gets a lot of influence over how the story goes; however, the author may end up effectively having to write several different games.

Examples: Katawa Shoujo; Magical Makeover.

Floating Modules. A mode only really possible in computer-based works. There is no tree – or, while there may be scattered twigs and branches, there’s no trunk. No central plot, no through-line: modular encounters become available to the player based largely on state, or perhaps randomly.

Effects: This is a challenging style to write for, both because it’s difficult to intuitively grasp – writers tend to rebound quickly to a more unified structure – and because few assumptions can ever be made about prior events. Without a large amount of content, the method tends to collapse into a linear system. Because play mechanics are largely about altering stats in order to negotiate a world, there’s a strong incentive to expose those stats to the player; repeated events chosen only to affect a stat (grinding) may be a feature.

There are different approaches to floating-module: Emily Short has a tentative categorisation here.

Examples: Pure examples of modular design are relatively rare. King of Chicago (hat-tip Sean Barrett) is an early example. StoryNexus and its conceptual relatives (e.g. Bee) inhabit this space, though they generally impose some more linear-progression structure on it. (Alexis Kennedy uses the term quality-based narrative to describe the general approach: ‘pieces of story like mosaic tiles, not pipes or complex machinery.’)

cyclegrowLoop and Grow.  The game has a central thread of some kind, which loops around, over and over, to the same point: but thanks to state-tracking, each time around new options may be unlocked and others closed off. This is a very general pattern, and can co-exist with many others. Trapped in Time, for instance, is basically a cycle-and-growing Gauntlet; Bee tames its floating-module nature with a year-long loop structure.

Effects: Loop and Grow emphasizes the regularity of the world while retaining narrative momentum. A justification is needed for why whole sections of narrative can repeat: the player-character is often following routine activities in a familiar space, engaged in time-travel, or performing tasks at a certain level of abstraction. This regularity often comes at the price of openness: many stories with a strong Loop and Grow structure involve a struggle against confinement or stagnation.

An important variation of loop-and-grow structures is spoke and hub: the game has several major branches, but they all originate at and return to a central node or set of nodes. The player may go out along each spoke once, or many times.

Examples: Bee, Trapped in Time, Solarium.

This entry was posted in cyoa, interactive fiction and tagged , , . Bookmark the permalink.

96 Responses to Standard Patterns in Choice-Based Games

  1. nothings2 says:

    “Floating modules”: An example that comes to mind is King of Chicago:

    • nothings2 says:

      In case you’re not willing to wade through it:

      King of Chicago divides its content up into 230 modulee, each of which is a small choice graph. Each module has several target state variables and values as preconditions, and it chooses which module to used based on a least-squares fit of the target variables.

      However, it’s not actually choosing from all 230; the modules are also divided up amongst several subplots, which are processed independently. (If I understand correctly, only one subplot has choices, and the other subplot are cutaway scenes to show what other people are doing). And there must be some other thing as well, because he says it’s usually only choosing from 10-20 modules at a time.

      • Yeah, I’m really pretty curious about the approach – how many modules would you need to avoid the least-square fit being overdetermined? Now I have urges to tinker with things.

  2. dfabulich says:

    I bet Floating Modules could be done with any form of randomness. I seem to recall some crunchy gamebooks with random encounters determined by dice roll.

    Choice of Robots has a Sorting Hat final chapter. (Hence, 300,000 words.)

    • I can kind of imagine a big card-catalogue kind of thing with a complicated code system on the tabs, yeah. Horrible to implement as a book, but I shouldn’t underestimate the lengths people will go to in order to implement fiddly programs as paper gamebooks.

      • dfabulich says:

        Eh, random encounters can be easier than that: just roll 1d100, turn to that numbered section. *shrug*

  3. Bruce Nesmith says:

    With most of our games we use the branch and bottleneck paradigm. It’s not unintentional. It’s a conscious choice to maximize the apparent player choice while still limiting the work to be done in handling branches. It also allows us to tell a story where we get to control the major beats, i.e. bottlenecks. In the cases where there are multiple endings, it is usually like you drew it, as the final choice.

  4. Pingback: Behind the Screen: Pattern Recognition

  5. Schedim says:

    Reblogged this on Schedim's Blog and commented:
    Thoughts on the structure of adventures, this is important to have in the back of the mind when you asseble a RPG session.
    And also a souce of inspiration.

  6. Pingback: Choose Your Own Autobiography (Neil Patrick Harris) | Emily Short's Interactive Storytelling

  7. Pingback: Interesting Links for 29-09-2015 | Made from Truth and Lies

  8. Excellent stuff there – thanks. 🙂

  9. 1998’s “Patterns of Hypertext” lists some of these and many other patterns that you may find interesting. It’s really neat how authors discover and re-discover and re-invent patterns, a little different each generation.

    • Ah, excellent; thank you. I don’t have much familiarity with the 90s-hypertext style, so this is very interesting (and there are definitely things in here which elicit WHY WOULD ANYONE POSSIBLY WANT TO DO THAT knee-jerks, which may or may not be a sign that learning is taking place.)

  10. Pingback: On Art And Video Games… Again | MultiPlatform

  11. Pingback: Windhammer Prize 2015: Isaac Newton: Badass Ninja Crimefighter (Stuart Lloyd) | Emily Short's Interactive Storytelling

  12. Pingback: CYOA Structures: Windhammer 2015 | These Heterogenous Tasks

  13. Pingback: Lifeline 2 (3 Minute Games) | Emily Short's Interactive Storytelling

  14. Pingback: Romance without routes. Breaking the CYOA/VN formula | Algorithmic Anthill

  15. Pingback: IF Comp 2015: Cape | These Heterogenous Tasks

  16. Pingback: Best Games Writing | Rock, Paper, Shotgun

  17. Pingback: The Sunday Papers |

  18. Pingback: Verbündelt – Der Mokkograd-Report | Mokkograd

  19. Pingback: Windhammer Prize 2015: After the Flag Fell (Felicity Banks) | Emily Short's Interactive Storytelling

  20. Pingback: IF Comp 2015: Map (Ade McT) | Emily Short's Interactive Storytelling

  21. Jacqui says:

    The Goosebumps CYOA books would start out as a gauntlet, then have a fairly obvious branching point (do you follow the mystery dude or do you just ignore them?) from which you have two other gauntlets that followed completely different (and mutually exclusive) stories. The later books started getting more experimentative with state tracking. (Rather memorably, one of the books asked if you got a rope, and if you said yes, the book would chastise you because there were no ropes anywhere in the book and would give you a THE END as punishment.)

    • Anti-cheating elements are pretty common in paper CYOA. My favourite is from Zork: The Forces of Krill:

      There are no Magic Sneakers and no Prince of Kaldorn in this book. You have been cheating. Vindictus, the Patron of Decision Novels, appears. Reaching out of the book, he casts a spell on you, and you turn into an unbelievably ugly toad.

      THE END
      Your score is negative fifty million billion zillion points. The score for the best ending probably isn’t important to a cheater like you who probably looks at the last page first.

  22. sl1605 says:

    Reblogged this on Lloyd of Gamebooks and commented:
    Getting to know the IF community better. This is an excellent post.

  23. Pingback: No Longer IF Comp 2015: Emily is Away | Emily Short's Interactive Storytelling

  24. Pingback: IF Comp 2015 Roundup | Emily Short's Interactive Storytelling

  25. Pingback: IF Comp 2015: Two Encounters in the Woods | Emily Short's Interactive Storytelling

  26. Pingback: Reports From the Field » Blog Archive » Research and future MMO Mechanics articles

  27. Pingback: Non-Linear Fiction | Design To Game

  28. Pingback: Scene chart, day one | Algorithmic Anthill

  29. Pingback: Creative Writing Career Ep7: Storytelling in Video Games | CREATIVE WRITING CAREER

  30. Pingback: CWC Podcast Ep7: Storytelling in Video Games | CREATIVE WRITING CAREER

  31. Pingback: CWC Podcast Ep10: Story Games and What Makes Great Story in Video Games | CREATIVE WRITING CAREER

  32. Pingback: My Name is Tara Sue | what will you do now?

  33. Pingback: Morality and Diegetic Agency in IF (via Radio K on Slouching Towards Bedlam) | Emily Short's Interactive Storytelling

  34. Pingback: Realities of Working in Video Games | Military Veterans in Creative Careers

  35. Pingback: GDC Highlights, Tuesday/Wednesday | Emily Short's Interactive Storytelling

  36. Pingback: Conference access, and related topics | Emily Short's Interactive Storytelling

  37. Pingback: Spring Thing 2016: Ms. Lojka, Tangaroa Deep, Sisters of Claro Largo | Emily Short's Interactive Storytelling

  38. Pingback: You Are An Online Clickbait Satirist. Can You Hack It In The IF World? | SPAG

  39. Pingback: Beyond Branching: Quality-Based, Salience-Based, and Waypoint Narrative Structures | Emily Short's Interactive Storytelling

  40. Pingback: Telling Stories in an MMO – Final Form!!! | Reports From the Field

  41. Pingback: Mark Bernstein on Hypertext Narrative | Emily Short's Interactive Storytelling

  42. Pingback: Design Pillars – Part III: Narrative Choice, & Consequence – Forged of Blood

  43. Pingback: I’m Your Forgotten Past: The Dubious History of Interactive Film | SPAG

  44. Pingback: Interactive Digital Narrative: Theory | Emily Short's Interactive Storytelling

  45. Pingback: Aviary Attorney (Sketchy Logic) | Emily Short's Interactive Storytelling

  46. Pingback: Narrative Structure in Operator – "SOME OF THIS IS TRUE AND SOME OF IT IS NOT."–ALLEN

  47. Pingback: IF Comp 2016: Black Rock City (Jim Munroe) | Hannah Powell-Smith

  48. Pingback: Elements in the Last Seven Cybertext of the IFComp | Far Far Futures

  49. Pingback: Finding Authentic Agency in Black Rock City – No Media Kings

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s