Branching narratives

Text & playables by Christina Lassheikki (@classheikki)

This experimental article is part of the Playable Concepts project at Aalto University.

Choices! In games! It’s not such a novel idea, though, is it? After all, we’ve had Choose-Your-Own-Adventure books since the 1970’s, and in interactive fiction, choice-based games and branching narratives have been a staple since, well, the start. Still, as a young narrative designer, it took me a while to wrap my head around what happens to a game’s narrative when we give the player a choice. Writing for branching narratives turned out to be tricky - and at the core, isn’t creating one good story hard enough? Isn't it hard enough to design one interesting path for players to take? Still, as a player I find myself much more invested in games that let me affect the direction of the narrative. Let’s take a peek at what branching means, why branching narratives matter, and why players like choices in games. This article builds on works and talks by other narrative designers and game writers, and also features small playable concepts to help explain the concepts and structures described in this text.

Playable concept 1 :
Example of a small branching narrative playable. The constellation chart acts as a map of the branching.
(Made using Twine - play on itch.io if the embed doesn't load: rainy paths on itch.io)

What do we mean by branching narratives?

In a nutshell, branching is a story design pattern game writers and narrative designers use in interactive narratives and games. It seems like there’s agreement that in a branching narrative, not all players will experience the same story. In this sense, branching is a narrative design tool that game writers can use to deliver their story, and in essence, narrative branching happens when you give your player a choice that impacts the story of the game.

During the ‘Developing Branching Narratives’ panel of LudoNarraCon 2020, the panelists Graham Reznick, Sam Maggs, Karin Weekes, Sam Barlow and Christian Divine discussed what branching narratives mean in games:

“[...] at a basic level, branching would be — [...] are you presented with the decision where you can choose at least one of two options. At its base level, you know — not just reading through a book. So anytime a player, or a viewer, takes an active part in deciding which [way] the plot’s going to go, for me, that’s what interactivity is.” (Karin Weekes, 2020)

As Karin Weekes (of BioWare) explains, branching in games happens when players are asked to make a decision, and presented with options. Sam Maggs continues on it this way:

“[...] any time a player makes a decision through gameplay, or any, any other aspect of the world. [...] You know, they’re making a choice between picking up item A or item B, and that affects something later. Or any time the narrative kind of splits, and could go one of two directions. Even if it comes back together at a certain point, that’s a branch.” (Sam Maggs, 2020)

Depending on how branching is used within the game, and structured, the player may see it as either an obstacle that hinders progression, minmaxable gameplay, a labyrinth or puzzle, or even as a reward. There are of course different ways to go with your narrative as well - a branching narrative is only one way to structure the story. On one end of the spectrum, expansive simulations can let players piece together meaning in a chaotic sandbox, and on the other end, even high-budget filmatic AAA games can have completely linear storytelling - and either could be engaging and immersive.

Branching narrative terminology

What are the tools in the shed of the branching narrative designer? Well, like much terminology in games, the vocabulary used for narrative design is varying. The longest tradition of analyzing choice-based structures can probably be found in the interactive fiction community, and designers like Emily Short and Sam Kabo Ashwell in particular have written in length about patterns. For ease of reading, here are some terms I’ve found useful.

Branching would be a point in which a story splits into two or more paths. At a point of branching, a story’s structure expands - like a new branch that grows out on a tree.

Now, if we give players three unique choices each time the game branches, and we follow through each of those choices with unique content, the amount of content required to make the game grows - literally - exponentially, as Carlton Reeve points out in his 2013 blog post. This is usually not feasible, nor actually desirable - after all, most players will play through the game only once. Simply put, we need ways to keep the amount of branches manageable. (Carlton Reeve, 2013)

Merging branches is when two (or more) branches come together once again. For merging branches, other words have been used as well; like collapsing branches, or converging branches. Carlton Reeve call these points shared nodes (Carlton Reeve, 2013).

Parallel storylines or parallel tracks would be a way to describe games that have key choices that lead players onto paths that end up merging later on. A similar concept would be what Emily Short and Sam Kabo Ashwell call a sorting hat structure. (Short, 2016)

Pruning would be a way to cut a branch short. This can either be through making it a dead end, from which the player or reader backtracks; through looping back to earlier, or by merging.

Conditional content through state checks and variables are then tools to make callbacks to player choices later on. For example, a player who chose to help a particular NPC earlier on might be greeted in a friendly manner, while one who stole their clothes might get attacked.

A simplified chart of the branching and merging in 'rainy paths' (see Playable concept 1 at the start of the article), and the structure as seen inside Twine.
In two of the three main paths, the reader is allowed a choice, but the ending isn't affected by the option chosen - causing a merging point.

Agency and meaningful choices

“How do you make choices matter? So - you don’t, is the answer, and that’s the proper essay answer - you don’t make choices matter. Because choices aren’t important. Choices are a tool for getting the player to do two things; firstly, to buy into the game, and be hooked by it, and secondly, to take some responsibility for it. So choices are intriguing, and choices are also condemning. And both of those are just ways to grab the player; get your hooks in them, say, ‘no, you’ve got to pay attention to me; you can’t look away from this’, while you do whatever it is you want to do in the first place.” (Jon Ingold, 2021)

In his 2021 WGGB talk ‘Making choices matter’, Jon Ingold claims, perhaps a bit provocatively, that choices themselves do not matter to players. A choice itself does not tell story, but it does give the player a reason to pay attention to the story being told.

Agency is often used as a term to describe the difference between games and non-interactive media - the obvious example being film. For example, Will Wright speaks about agency versus empathy in his game design class on MasterClass as the big difference between stories in games and stories in non-interactive media; the argument being, that a film relies on empathy for story engagement, while a game relies on agency. But we could also argue that the only true agency a player is given, is the choice not to keep playing - since every single affordance of a game is built. And so it’s perhaps more accurate then, to the craft of game storytelling, to create an illusion of agency through offering an adequate amount of choices with reasonable consequences. Even in cases where the player does make choices that affect the story, that effect doesn’t have to be the one the player wants or hopes it to be. And so, we can question the role of player agency - overall - in our branching. We are used to the idea of building games as power fantasies, in which the player through their actions affect the world around them in positive ways. Plans fail, misinterpretations happen, and the road to hell is paved with good intentions - and stories in games can absolutely reflect that - a good story doesn’t mean the player always succeeds. Cat Manning also points out that not all choices need to have mechanical effects - choices that let the player express their feelings on a subject, or reflect on or validate their experience, can be just as impactful. (Manning, 2018)

Still, by meaningful choices in game design we often tend to default to choices that a) have an effect on the game, that the player sees as connected to the choice they made, b) the player, as they make that choice, understand that they are making a choice, and c) the player, as they make the choice, have enough information to distinguish between the choices and has an expectation - however flawed - of what the outcome of the choice may be. We could call choices that fill these requirements conscious and transparent; after all, they’re choices players know they are making, and that they can see the effects of. But, branching could just as well happen behind the scenes, under the hood of the game. As, Sam Barlow points out during the LudoNarracon Panel Developing branching narrative:

“I’d say, a branching narrative is one where it’s possible for two players to play the game, and in describing the events of the game, have meaningfully different stories to tell.” … “The code and the — in representing me as the author — was making choices for them, so everyone got this unique and different narrative, different plot, different moments, but there wasn’t a conscious choice. Which I think for me is a useful way of differentiating something, because I think often when people talk about branching narratives, it’s more the choices, that they’re thinking about, than necessarily even like the structure of the narrative or how it might branch.” (Sam Barlow, 2020)

The nature of the choice interface can also be such that players don’t always feel as if they were given a choice. In games like The Dark Pictures Anthology: Man of Medan (Supermassive Games, 2020), and Heavy Rain (Quantic Dream, 2010), players have to pass quicktime events - essentially controller challenges - at key points of the story in order to keep specific characters alive. In Man of Medan, for example, an early-game series of events can either lead to a character’s early death or their daring escape - but regardless, the story continues, with or without the character.

Telegraph choice outcome - or don’t

How can a narrative designer then build an idea of the outcome without stating it outright? Telegraphing is a term that refers to signalling what is to come. It’s perhaps most commonly used to refer to an animation that plays before an enemy makes a move in combat, as a variant of the principle of anticipation, as laid out by Frank Thomas and Ollie Johnston in the Illusion of Life (1981). Telegraphing is also used in game writing, to refer to the, often shortened or recapping, versions of options the player is presented with. Telegraphing can also use visuality; like icons. For example, the player character Hawke in Dragon Age II (BioWare, 2011) usually has three options in his or her dialogue wheel; a diplomatic or helpful one (icons: an olive branch, a halo and wings - colors: green and blue), a humorous or charming one (icon: a theatrical mask, a diamond - color: purple), and a direct or aggressive one (icon: a fist, a gavel - color: red) (Fandom.com). But, the telegraphing can also be more subtle, like the color coding of dialogue options in Night School Studio's Oxenfree (2016), which is never explained and keeps the player guessing, but gives an intuition of slight differences in tone and helps visually differentiate the options.

Telegraphing choices can also be criticized as unnatural, and predictive. After all, if the player knows what is going to happen, part of the thrill might be gone. This is closely related to how stories can create suspense - we present the player with a situation, from which they can see multiple outcomes but they hope for an intended one. Where in linear storytelling moments of suspense are resolved by the author, in interactive narratives, players resolve them - using, of course, whichever tools the narrative designer has given them.

Regardless of what form the telegraphing takes, we could say it can aim to be transparent - representative of what the character will say or do, or the outcome of the choice will be, or suspenseful - create suspense at what the character might do, or the outcome might be. One or the other isn’t necessarily better.

Micro level and macro level branching

When we’re making a branching narrative, we can of course also distinguish between moment-to-moment decisions - such as a dialogue choice - and the overarching plot structure - say, choosing which faction to become your ally. We can perhaps distinguish these as micro level branching and macro level branching, depending on how far into the game the consequences carry and with what impact. Sam Kabo Ashwell has made a wonderful overview of patterns in choice-based games (Ashwell, 2015), which for example Emily Short has built upon with the introduction of storylet structures, and in her blog she also explains structures that gives examples of alternative structures to branching (Short, 2019) and small-scale structures (Short, 2016).

During the ‘Developing Branching Narratives’ panel of LudoNarraCon 2020, the panelists suggested a concept from linear screenwriting - the A-plot and B-plot - as another way of seeing this. As Sam Barlow points out, a game with a save the world story is most likely going to end with the player saving the world. How do narrative designers then make the story matter, if the outcome is clear to the player from the start? During the panel, Karin Weekes explained the BioWare approach. In essence, it’s to let players explore complex character relationships in-depth, while the main plot remains less branching. Choices made in relation to the main plot affect the character relationships, but the reverse is less prominently used. In for example the fantasy banter RPG Dragon Age: Inquisition (BioWare, 2014), the player makes choices on which companion characters to bring along for the events of the main plot. During the main quest sequence Wicked Hearts and Wicked Minds, a sequence of court intrigue and high-stakes political machinations in the midst of peace talks during a civil war, these companions have unique comments and a presence throughout much of the level. If the player is involved in a romance with one of the characters, they get an optional cutscene with their paramour, the contents of which are determined by the fallout of the main plot.

Another example would be what Heussner et al. describe as multilevel storytelling, combining linear and nonlinear storytelling. For example, an overarching, linear world narrative with smaller, player choice-driven stories. They suggest this structure works especially well for MMORPGs and other multiplayer games. (Heussner et al. 2015)

Sam Kabo Ashwell introduced the concept of gauntlet structure, meaning a fairly linear path with branches that are pruned quickly - through failstates, backtracking, or rejoining the central path - this could either be on a macro level (game wide) or on a micro level (say, one conversation in the game) (Ashwell, 2015).

In other words, we can see the structure of a game’s narrative as fractals; or russian dolls. Once again it comes down to scope, budget and project needs - there’s no one structure that’s better than others, and no rule that is absolute.

Well. Maybe there’s one exception…

Designing choices

According to Jon Ingold:

"Incidentally, three choices is the correct number of choices. [...] If you have two choices, it’s binary, and people feel there’s a right one, and a wrong one. If you have four choices, it looks like you don’t know what you’re doing, and the scene, it lacks focus. It’s too much for people to process. Three is the correct number." (Ingold, 2018, [26:26])

The reasons he give are, at the core, quite simple: as designers, we want to give a breadth of choice that covers an adequate amount of reasonable things players would want to do in a given situation, without inducing analysis or choice paralysis due to giving too many options. Binary choices, as he explains, are dangerous, since they quite easily end up as a right and a wrong choice - not necessarily a moral dichotomy, but, rather, a choice that the writer sees as the true choice and then the other option that kind of just needs to be there for the sake of choice - leading to a false choice. Binary choices should therefore be used sparingly for effect, if at all.

Well, if the number of choices is three, then what should they be? In his 2018 talk, Ingold also gives a simple structure to use when designing these three choices - accept, reject, deflect - which can be as simple as “Okay”, “No”, and “What about this?” as the options to any given situation - but can also allow a designer to approach writing a dialogue as one would a battle sequence - for example, accept as avoid confrontation, reject as attack, and deflect as parry. (Ingold, 2018)

(If you paid attention to Playable concept 2 - The Toothless Dragon Tavern - earlier, you might have noticed each option was marked with R, A or D - indeed, Reject, Accept and Deflect was used for the choice design.)

The subject of what choices to give players and how they are framed is what Mahowter et al. call choice poetics. A false choice is an example of what Mahowter et al. call a choice idiom, of which they list quite a few, including the unchoice - a situation framed like a choice, but with only one option. (Mahowter et al., 2014)

If at their best choices make players feel responsibility for their actions, then we of course also need to consider what the ethical implications of the choices we offer are. According to Sam Maggs, referring to BioWare games, the vast majority of players want to just be nice if presented with that option (Sam Maggs, 2020). Research also suggests the same; players on the whole go for heroic or morally “good” alternatives, although players who play a game multiple times may choose to take a more evil route on their second playthrough (Lange, 2014).

In designing the choices, it's then good to be aware that the average player tend to go for morally "good" alternatives if they are available, and design around this behavior. If a writer really wants to frame the story as good versus evil, then perhaps the player can be tempted to commit heinous things with mechanical boons, or perhaps good intentions can be twisted. But, more importantly, I think we should always ask ourselves what worldview our choice space is reflecting. If good intentions are always punished, or if helping people is going to cost you a chance to level up, what bigger story are you telling the player through the sum of these outcomes?
One method to move beyond the good/evil dichotomy that Heussner et. al mention in The Game Narrative Toolbox is to explore a theme through your story. The example they give in from Dragon Age II (BioWare, 2011), in which stories are derived from the theme of security versus freedom. (Heussner et al. 2015) Many, if not most, of the choices - both on micro and macro level - in the game are variants on that theme.

This is also in line with how Karin Weekes described the process of approaching character writing at BioWare:

And the other thing that we do, is for every game, and every major mission within that game, we’ll have themes. And so we try to keep those discussions; it’s not right or wrong, it’s exploring faith. There’s a myriad different ways for people to respond to faith. (Karin Weekes, 2020)

This approach has many benefits, and it can really help characters come alive. Of course, since people have different values, some themes will not feel like a source of dilemma or depth to all players.

Choice outcomes

In addition to narrative outcomes or gameplay effects, choices can have structural effects on the branching itself. Here are just four; the failstate, moving on, the trapdoor, and the loop.

A classic from the Choose-Your-Own-Adventure haydays would be the failstate. Open the wrong dungeon door? Monster - you’re dead, return to the start. A lot of players find these frustrating in narrative-driven games since they mean you’ve chosen, simply put, wrong.

Playable concept 3:
Failstates can feel quite frustrating. (Made using Twine - use this link if the embed doesn't load: failstates)

Graham Reznick discussed this as follows during the LudoNarraCon panel Developing Branching Narratives:

“Did you fall into a pit, did you run into a monster, did you die, you know, and then you start over and you progress again until you pass all those failstates. But that doesn’t really happen in narrative. It kind of can’t happen in narrative. Everything has to be moving forward, and everything has to add up to some amount of meaning. Whether that’s a very curated meaning, or not, it still has to add up to that meaning in the player’s head. So hitting those failstates will immediately pull them out of the world of the story, in a way that you kind of never experience in a linear traditional narrative.”
(Graham Reznick, 2020)

Player death and redoing a level is still a really common design pattern in games, but as Reznick discusses, the experience is jarring. We could say, simply, that it causes ludonarrative dissonance, unless repeated death and resurrection is addressed and explained in the story.

And, as Sam Maggs put it later in the panel:

“I feel like videogames has like, we have 30, 30 years of experience, in this industry if not more, of like, ‘what are the other kinds of consequences for your actions that [...] you can have’ — be they small, or huge, or world-shattering, or whatever, beyond ‘you died’.” (Sam Maggs, 2020)

For example, in the greek mythology-inspired roguelite game Hades (Supergiant Games, 2020), player death is contextualized as attempts to escape the Underworld, and a central driver of story in the game. According to a tweet by the developers, Hypnos - an NPC in the game - even has dialogue lines for 75 unique death scenarios (Supergiant Games, 2020 [tweet]).

In contemporary game design, limiting attempts - and thus lives - is quite rare. Game Maker’s Toolkit has a very informative video that discusses the subject in length. As they point out, Crash Bandicoot 4 offers a retro and a modern mode, in which the main difference is whether lives get depleted or not upon player death. (Brown / Game Maker’s Toolkit, 2020)

Failstates obviously do not have to lead to death in a game, but, a reason they do is to create high stakes (it could be argued, of course, that a constant threat effectively undermines the stakes of life-or-death situations plot points), but perhaps more, to avoid the player frustration of backtracking from a lengthy dead end.

Avoiding backtracking leads us to the perhaps simplest choice outcome, which would be moving on - or simply put, following through with the player's choice. The player made a decision; the game reacts accordingly; no going back. We could imagine moving on to be the default outcome of a choice inside the choice structure, but that’s not always the case.

The loop and the trapdoor are two more structural elements that Jon Ingold explains in his 2018 AdventureX masterclass titled ‘Sparkling Dialogue’. The game might loop back to the choice. A classic type of loop is what Emily Short calls the re-enterable conversation node; imagine talking to an NPC, and being given the options to discuss a variety of subjects (Short, 2016). But, as Ingold points out, loops hinder progress, and can therefore feel frustrating to players - which can be used to great effect as a means to punish players. A trapdoor, on the other hand, is a way to move on and skip a part of branching content; think of it as an invisible shortcut. (Ingold, 2018)

The structure of your branching, as Jon Ingold points out, absolutely does carry meaning as well. (Ingold, 2021) A structural element, such as a trapdoor or loop, could feel like a welcome shortcut, or punishing, depending on the context. If you, say, are able to avoid an argument all together through navigating a dialogue tree cunningly, a player may feel triumph at sidestepping the conflict. Or, perhaps, the player was paying attention earlier in the game, and is able to convince a companion character that they do care about them, thus avoiding unnecessary heartbreak. Of course, the opposite could be the case as well - a trapdoor as a punishment. We can imagine a flirtatious or intimate scene, where the wrong move makes your paramour lose interest, or a high-stakes interrogation, in which tipping your hand could make your informant clam up. Similarly, a loop can either feel punishing - hitting dead ends - but on the other hand, loops might alleviate a fear of missing out in a situation where multiple options seem tempting.

Callbacks

As we've already established, creating unique content for every single decision a player makes during a game isn't very practical - but still, it's a nice feeling to have your earlier decisions matter. One of the neatest tricks we can pull on a player is the callback; another storytelling trick (used widely in noninteractive media as well as in improvisational storytelling) to create a sense of cohesion. In games where text is presented to players callbacks can be really easy to implement, while voiced games can be limited by budget. Still, it's a neat way to keep branching content manageable, but still remind the player that the choice they made was recorded and that their actions do have consequences - even if they're small.

References:

Ashwell, Sam Kabo . Standard Patterns in Choice-Based Games. These Heterogenous Tasks (blog), posted on January 26, 2015. Accessed January 29th 2021.

https://heterogenoustasks.wordpress.com/2015/01/26/standard-patterns-in-choice-based-games/

Brown, Mark: “Are Lives Outdated Game Design?” Game Maker’s Toolkit. Uploaded on 15 Oct 2020. Watch on YouTube: https://www.youtube.com/watch?v=c2CLO8CcBjg

Fandom.com: Dialogue wheel (Dragon Age II) . Accesssed 7 March 2021.
https://dragonage.fandom.com/wiki/Dialogue_wheel_(Dragon_Age_II).

Heussner, Tobias, Finley, Toiya K., Hepley, Jennifer and Lemay, Ann. The Game Narrative Toolbox. 1st ed. London: Routledge, 2015. (E-Book)

Ingold, Jon: Sparkling Dialogue: A Masterclass - AdvX 2018. Watch on YouTube: https://youtu.be/_vRfNtvFVRo

Ingold, Jon: Making choices matter in videogames. Talk for Writer’s Guild of Great Britain. Posted 16 Feb 2021. Watch on YouTube: https://youtu.be/gHIBzdVpfNw

Lange, Amanda: "You're Just Gonna Be Nice": How Players Engage with Moral Choice Systems

Journal of Games Criticism, 2014. Accessed Feb 18th 2021. Available online:

http://gamescriticism.org/articles/lange-1-1/

Mawhorter, Peter, et al. "Towards a theory of choice poetics." FDG 2014. (2014). Accessed Feb 18th 2021. Available online:
http://www.fdg2014.org/papers/fdg2014_paper_19.pdf

Manning, Cat: Successful Reflective Choices in Interactive Narrative. Posted June 19th, 2018. Accessed Feb 18th 2021.

https://catacalypto.wordpress.com/2018/06/19/successful-reflective-choices-in-interactive-narrative/

Reeve, Carlton: Exploring Interactive Narrative - Branching. play with learning (blog).

Posted on 21 October 2010. Accessed February 2nd 2021.

http://playwithlearning.com/2010/10/21/exploring-interactive-narrative-part-2/

Reznick, Graham (moderator), Maggs, Sam, Weekes, Karin, Barlow, Sam and Divine, Christian (Panel) Developing Branching Narratives - LudoNarraCon 2020 Panel, April 24-27 2020.

Watch on YouTube: https://youtu.be/mgD81pQlu1o Accessed January 29th 2021.

Short, Emily. "Beyond branching: quality-based, salience-based, and waypoint narrative structures." Emily short’s interactive storytelling (2016). https://emshort.blog/2016/04/12/beyond-branching-quality-based-and-salience-based-narrative-structures/

Short, Emily. “Small-Scale Structures in CYOA” Emily short’s interactive storytelling (2016).

https://emshort.blog/2016/11/05/small-scale-structures-in-cyoa/

Short, Emily. “Storylets: You Want Them” Emily short’s interactive storytelling (2019).
https://emshort.blog/2019/11/29/storylets-you-want-them/

Supergiant Games. @SupergiantGames (December 29th, 2020) “As a matter of fact, we have this handy little infographic about it, since the question's come up a few times! The numbers may not be 100% accurate depending on how you count but should be close.” [Graph with title ‘Voiceover in HADES’] Tweet (reply).

https://twitter.com/SupergiantGames/status/1343986323007238144 . Accessed March 4th, 2021.

Wright, Will,: Will Wright Teaches Game Design and Theory - Part 4: The Relationship Between Story and Games, Master Class. (accessed 2020)

https://www.masterclass.com/classes/will-wright-teaches-game-design-and-theory

Games mentioned in this text:

BioWare: Dragon Age: Inquisition. Released November 18, 2014. Played on PS4.

BioWare: Dragon Age II. Released March 8, 2011. Played on MacOS.

Night School Studio: Oxenfree. Released January 15, 2016. Played on MacOS.

Quantic Dream: Heavy Rain. Released February 23, 2010. Played on PS4.

Supergiant Games: Hades. Released September 17, 2020 (Early access released December 6, 2018). Played on MacOS.

Supermassive Games: The Dark Pictures Anthology: Man of Medan. Released August 30, 2019. Played on PS4.