The game narrative triangle


There doesn’t seem to be a lot going on in the game scene right now (at least for people not interested in motion controls, ever-more-unsettling modern war shooters and misjudged forum policies), so I’m taking advantage of the quiet time to go back some of the basic structures in games. Today I’m taking a first-principles look at the kinds of stories videogames tell.

It’s generally agreed that there are two types of game stories: what the script says and what the player does. Or as Valve writer Erik Wolpaw put it in a presentation at the Game Developers Conference in 2008:

Games tell two stories: the story story – the narrative story that’s the sum total of a game’s cutscenes and dialogue; and the gameplay story – the story described by the actions the player takes in the game world.

Game designer Marc LeBlanc proposed the names “embedded narrative”, for the story set into the game by the designers, and “emergent narrative”, for the story that emerges from the process of playing.

There’s a third type of story in videogames, but I’ll get to that in a minute.

The distinction between embedded and emergent narrative can get a little blurry. No game is entirely composed of one or the other; every game contains at least one embedded story (even if it’s as minimal and abstract as Tetris) and at least two potential emergent stories (in the simplest kind of game, a coin toss: heads or tails, win or lose).

Some atomic elements of game narrative can be definitively classified as embedded or emergent. Individual cut scenes and scripted dialogue are purely embedded, as the player has no control over them. Direct communication between players in a multiplayer match is purely emergent, as the game creator has no control over it. However, most events fall somewhere in between: the actions of a player’s avatar, for example, are directed by the player but performed in a framework of rules, animations, sound effects and story context designed by the game’s creators.

We could represent this with a scale:

This scale is missing one factor, however. Videogame narrative elements come from three sources: the game designer, the player and the computer. In a complex system, the game software will throw up events or images the designers didn’t directly create, and may not have been able to predict. This is another form of emergent narrative, distinct from what the player or designer puts into the game.

Take an open world game for example. If you were to describe “the story of Grand Theft Auto IV”, you would talk about Niko Bellic, his dark past, his search for the American dream and redemption, his rise through the underworld hierarchies of Liberty City and his ultimate disillusionment. The game tells this story with cut-scenes that interrupt the gameplay and scripted dialogue that plays over the top of the gameplay. There are other stories that play out over the radio, some related to you, some not. These are embedded narrative elements, directly created by the game designer.

If you were to describe “what you do in Grand Theft Auto IV”, you would talk about stealing cars, speeding through crowded intersections, running from the police, jumping from buildings, making squealing handbrake turns and gunning down opponents. These are emergent narrative elements, created by the player.

If you were to describe “what happens in Liberty City”, aside from the havoc you wreak directly as the player and the scripted story elements told in cut scenes, you would talk about pedestrians walking beside roads, garbage trucks cruising the streets early in the morning, police officers patrolling, cars stopping at traffic lights and negotiating corners – civilians fleeing your gunshots in panic, police officers chasing your car at high speeds, gangsters pulling guns as you approach and cars smashing into other cars in panic as the traffic flow is churned into chaos. All of these things are controlled by the computer. In one sense, it is all embedded behaviour, as the rules and AI that govern when and how each action will occur are coded in advance by the game’s designers; however, when the moment-to-moment outcomes of that behaviour could not be predicted by the creators, they are emergent narrative elements, effectively created by the computer.

Now the game narrative scale looks like this:

In a complex narrative game like GTAIV, almost nothing is purely one kind of narrative or another. Anything that could be considered a spoiler is almost certainly embedded; cut-scenes, for example, are purely embedded narrative (unless you consider the player’s experience of watching them more broadly, taking into account their context, knowledge and emotional state, in which case they have a shade of player-created narrative to them). A more interactive experience, like a police pursuit, is a fluid concoction of embedded, player-created and procedural elements, each reacting to the others and shifting from moment to moment.

In practice, only very specific narrative elements can be precisely pinned down to a point on this narrative triangle.  Nevertheless, I think it’s a useful schema for considering the status of narrative elements in games. If nothing else, it’s a good shorthand illustration for one of the reasons game narratives are so much more complex than they outwardly appear.

If you have a different idea, of course I’d like to hear it. Perhaps a narrative hexagon or a four-dimensional story hypercube?  Or is there a game that just doesn’t fit this model at all?

Related posts:

  1. In defence of cut-scenes
  2. Narrative Excellence: Ultima VI
  3. A beginner’s guide to game genres

Tags: , ,

| More

Fraser Allison

Fraser comes from a long line of tinkerers and troublemakers, and the apple didn't fall far from the tree. He's an internet addict and a friend to animals. In 2010, he completed an honours thesis entitled "The prosthetic imagination: immersion in Mirror's Edge", which you can view here. You can follow Fraser on Twitter, or hang out at his house and play Top Spin, whatever.

More posts by
Email Fraser.


  1. Terrific clear thinking, well done. Very inspiring.

    It makes me wonder if veering too far towards each corner of the triangle might be a potential weakness in any design. Too much embedded content at the expense of other elements is un-interactive, too much procedural leads to over-complex or luck-based rule systems, too much emergent behaviour without content or rules can lead to an aimless experience. A bit of a one-dimensional rule of thumb…

  2. Lovely, lovely read. Never thought about AI like that before, but I suppose it’s been omni-present in most genres really, and will only become more prevalent as AI continues to improve.

    Best example I can think of – from GTA actually: a guy pulled up to a curb to hire a hooker but he had a passenger with him. The hooker tried to get into the car but to do so had to pull the passenger out, who in turn saw this as an act of aggression so started a fight. The hooker lost the fight and an ambulance turned up. Then I stole the ambulance. Just a beautiful line of the game procedurally generating a scenario and then me ‘reacting’ to it.

  3. Craig: I love the GTA vignette. I’ve never noticed anything quite so detailed play out while I was there, but the individual unscripted interactions between NPCs really help to establish the illusion of a city that would carry on when the player isn’t around.

    Alex: The rule of thumb is an interesting one. It reminds me of the Immersive Sim genre, which Rock Paper Shotgun has recently been doing a eulogy for; those were the games that tried to push all three types of story equally strongly by creating detailed, living worlds with a high level and wide variety of player agency. If you had unlimited money to get all your favourite game developers together to make the perfect game, that would surely be the path you’d have to take.

    It’s a handy rough guide, although I think the relative value of each type of narrative would vary for different types of games. I imagine you’ll always get less out of an embedded narrative in a racing game than an adventure game – or maybe I just haven’t played a memorable enough scripted racing game yet.

  4. This is a really interesting way to look at game narratives. Splitting the non-player-centric narratives up between embedded narratives AND procedural narratives is such a small yet significant step. Adding that third point makes it much easier to evaluate exactly what kinds of narratives are going on at any one time.

    Great concept and concise writing. :)

  5. Does it occur to anyone else that the parameters of Creator, Player and Procedural have the potential to be applied not just to games but to the human existential dilemma?

    Contained within the same triangle are Free Will, Destiny, Pre-determination and the quest for meaning.

    The very appeal of games vs. reality is the interaction with the creative vision of the games creator. While the role of “the creator” for most in real life is opaque, ephemeral or perhaps illusionary, the game creator makes clear the tasks most meaningful and rewards the player in simple non-subjective ways.

    It is because games interact with a mortal creator in an imaginary place that games are fun and play. If one considers them self engaged with moral reality, actions become work and important.

    Those who posses faith in a religion or deity derive meaning from their actions as though life is reduced to a game of moral actions with clear rewards. While those who experience disillusionment or uncertainty wrestle with motivation and meaning.

    In the same way, a totally open ended “choose your own adventure” game provides less motivation than one with a narrative path, a goal or existential threat. If there was no plot to advance and no experience to be gained, why fight the monster. If there is nothing to be bought, why earn money.

    Question: If a game was made “real instead of imaginary” could it still be fun to play? Can a game be made that helps clean up oil or distribute aid? More over, can a game be made that concerns the existential dilemma of human existence in such a way as to evoke spiritual evolution in the real world?

  6. JontyPTraveler: Interesting questions.

    There are games that at least educate players about preventing oil spills and distributing aid, and indeed the challenges of living in poverty. Few directly contribute to solutions, though.

    Google Image Labeler is one example, although the problem it solves is small: improving image search results by turning image tagging into a two-player mind-reading game.

    The question of how the structure of a game influences the meanings a player draws from it is something I’m investigating for a thesis at the moment, but it’s a tough sucker of a question. It seems to be true that our perception of our own agency within a game-world substantially affects how we feel about our actions in that world. I’d guess the same applies to our real lives, although if free will is an illusion, it’s a convincing enough one that it’s hard to shake.

  7. Yes, there are games that educate. In fact, all games educate, it’s just that what they teach isn’t always easily applicable to the real world.

    I think the SETI program had a semi-game like approach to crowd sourcing processing power for some kind of analysis of interstellar background noise…

    There is a bit of irony in the fact that gaming skills translate very effectively to a task like piloting the Predator Drones so effectively assassinating Taliban leaders and scores of civilians along with them.

    The question of how the structure of a game influences the meanings a player draws from it is an excellent subject for a thesis. Owing to my earlier postulation on the similarities between the real world, the game world and how games seek to limit the existential dilemma by limiting uncertainty, I’d like to point to the games we play in the real world, the structure of which intend to influence the meanings a “player” draws.

    For instance, in the 1300′s the great cathedrals of Europe presented a structured experience which for the commoner of the day was a multi-media overload. Glimmering stained glass stories backed by echoing resonant choral sounds elicited a mental state open to religious transformation and copious tithing.

    What is so different about the intro sequence in a contemporary game? Imagine the thrill quenched after 12 years of weighting for Starcraft 2 when the gamer finally watches the first cut scene with surround sound speakers blasting. Amen…

    I was involved with an “alternative reality game” called World Without Oil ( which invited players to investigate a plot loosely posted on any and all social networking sites. I was hired to make a few of the initial videos. What seemed to happen though was that the narrative plot the game designer began with became less important than the crowd sourced narrative.

    Perhaps somewhere there is the essence of how to create meaning.

    Imagine the scene from the novel “All is Quiet on the Western Front” when at Christmas the Germans and British cross no-mans land to exchange gifts.
    The structure of “the game” of the first world war, all the propaganda and nationalism broke down and was taken over by an impulse for comradery. From what I have read on the subject, that same scene played out in actually several times during the first world war.

    Like wise, not many videos of in game experience have equaled the epic Leroy Jenkins World of Warcraft incident. If you don’t know what I’m talking about, you have to look it up on youtube.

    When you write your thesis, I hope you include something about how to structure a game so that the players will break the structure.
    Don’t tell god I said that, because the real world ramifications are a little more interesting…

  8. Just had a laugh… I see Jeroy Jenkins is in your triangle… ; )

  9. [...] by Fraser Allison that’s made me want to spew some words on a bit of the subject. In The Game Narrative Triangle, Allison talks about the three aspects that actually make up a modern game. Emergent gameplay and [...]

  10. Thought provoking article. I have to say I was sceptical before reading and unfortunately am not quite convinced about the computer being an equal entity in this triangle yet. I still see computer narrative (as you recognise as well) as embedded narrative.

    It is part of the designed narrative space. Indeed this often includes random factors, but even their range of values is designed. If there is a third pillar acounting for uncontroller for but non-user generated narrative, I would say it is ‘other players’.

    As to the ‘can real problem solving be fun’ question: Yes, if the there are no concequences for the player, or at least he or she is not aware of them.

  11. Interesting article, but I think you may be ignoring the biggest factor in emergent/scripted narrative, which is that there’s a near-unbreakable wall between the two.

    My model of choice is the dualism of simulation & narration, understood in terms of player agency. The player ACTS in the simulated world, and OBSERVES the narrated world.

    While your model has some value, it divides the player’s participation into acting (in the simulation), observing (the narration), and witnessing(?) the emergent behaviour from the simulation, so it would seem to me not to be a triangle, so much as a deepening of one side of the simulation/narration dualism.

    There are (to date, at least) very few games which break down this wall. In Animal Crossing, there are narrative elements embedded in the simulation, which are allowed to emerge through player agency. Left 4 Dead’s Director AI shows promise as an embrionic form of dynamic Narrator. If there are others, I cannot think what they are (neither Facade, nor Storytron are not really suitable for embedding into game systems).

    For the record, I’m working on my own dynamic narrative system, which hopefully I will be able to demo next year.

  12. Yazveru:

    “Other players” as a source of narrative is an interesting suggestion. It shows that you’ve approached the issue with a different assumption about how the game is being viewed than mine. In game terms, yours is a first-person model and mine is an overhead view; you’re thinking about how the game looks from within as a player, whereas I’m thinking about how the game looks from the outside. I wasn’t conscious of my own perspective, but I think both are valid.

    Computer-generated narrative is an entirely distinct form of narrative in the abstract. However, it’s telling that there is no example of pure computer narrative mentioned in the article. The only examples that we can point to in existing games are partial cases, including the examples I give. It’s difficult to imagine how a pure computer narrative could come about without input from a human designer (that’s why glitches are useful examples, as they are at least unintended by the designer). However, as an element of game narrative becomes more complex and less predictable, more dependent on the interaction of computer processes and less on direct human input, it moves further and further away from the creator’s design; the logical, if impossible, endpoint of that is the pure computer narrative.

    While this might seem like a technical disqualification, it’s important to remember that the other two points of the triangle are equally impossible to find in a pure form. Even in a purely pre-rendered cut scene, the computer is at least running the video file; the player is at least interpreting what they see and hear.

    Dan Stubbs:

    It’s great to hear from someone working to crack the actual problem of dynamic narrative; all this theorising won’t change any games by itself!

    I’m interested to know why you say there’s an unbreakable wall between emergent and scripted narrative. If I read you correctly, you don’t mean that there can be no true hybrid of emergent and scripted narrative, but rather that they refer to different planes of the mental experience: scripted narrative is the stimulus players respond to, and emergence is the possibility space they play within. Is that about right?

    I’m at odds with your description of the gap between types of narrative as a wall or divide, which seems to be due to a difference in the fundamental way we’re thinking about the concepts. Let me back up a bit to explain this.

    I have a particular bias against the idea that things exist in a pure form (Plato is my least favourite philosopher); the common tendency to divide and categorise strikes me as a result of the way the human brain is wired to label ideas neatly in order to process them more easily. (This model of processing, by the way, is explored exhaustively in Ian Bogost’s rather good book Unit Operations.)

    This resistance to discrete ideas is an overlooked factor in the triangle article. A few people have mentioned that this article takes one step away from the traditional model of emergent/embedded stories, but it actually take two steps. The first step, which is often overlooked, is that I’ve represented embedded and emergent narrative on a scale, rather than as two separate points or as a Venn diagram. In other words, I don’t think of it as a dualism, the way you’ve described. By the same token, to the extent that the triangle should be taken seriously at all, it’s intended to be read as an area in which game narrative elements can be roughly located, rather than three isolated points. As I mentioned above, I don’t think any of three narrative types exists in a pure form.

    Bringing it back to your point: I see the player’s participation within a game as a fluid conflux of observation, interpretation and action – not as multiple distinct processes but as aspects of an integrated behavioural process.

    (I’m trying to use plain English here, but I’m not smart enough to make this idea sound as simple as it really is!)

    I admit this mashing together of apparently discrete concepts may be an unhelpful approach to solving the practical problems of dynamic narrative. On the other hand, the subconscious tendency to compartmentalise aesthetics, gameplay, narrative and so on is part of the reason game stories are so frequently disjointed and unfocused, so perhaps it’s worthwhile.

  13. Austin Ivansmith

    Hey Fraser,

    Overall I really liked your article. I feel your theory is very good, but not quite 100% there. I have shared my thoughts on Kotaku, which I am sure I will get flamed for very harshly, and thought I would venture over here and share as well.

    “I like this but think the “computer” aspect is really up for debate and further discussion. The examples called by Mr. Allison are all bugs within the games, but his description on “procedural” leaves a more open ended ideal of AI within the game. It is also a little insulting (just a little, and unintentionally) to designers and programmers because it presupposes that these little narratives are not the intention of the creator, and it is an attack on the creativity and intelligence on the creative team.

    I wonder if the “procedural/computer” part of the triangle shouldn’t be there because they are still experiences outside of the embedded path. The player encounters these events and gathers their own interpretation from them, but in Mr. Allison’s own descrption the emergent experience still falls within a set of rules and many of these rules are dictated by and create these “computer” experiences as well.

    Or if it should just not be considered “emergent” and instead should fall under a new type of category of bugs which were not intended by the creators and not initiated by the player.”

    Of course I wrote all this before reading about Craigs experience in your comments. Hooray for me and jumping the gun. I guess for me I had always felt that anything outside of the story was the single emergent experience regardless of what initiated it, whether it be computer or player. Or maybe I just don’t like having two categories (emergent, embedded) breaking up a triangle; feels a bit unbalanced. In any case, I think it is a good analysis and I look forward to seeing where you take it from here.

  14. I think your model of thinking about narrative would probably be helpful for designers, developers etc. But I almost want to drop the word narrative to push your thesis a little bit. Rather than speaking about embedded and emergent narratives, why not talk about drama and sport. Drama is the story we are told. It has rules and conventions which we have no control over (other than to turn it off). Sport is virtualized violence carried out against an opponent (real or AI) according to an agreed upon set of rules, whether it is football, chess or Call of Duty. Of course drama is interactive in the sense that – well I hope I don’t have to explain that. Sport is full of drama, and there are narratives that we tell about sport. However we have yet to really confuse the two. (I would tend to put unusual AI interactions into the sport category (such as when the ball takes a crazy bounce because it’s wet) which doesn’t mean they aren’t fun or entertaining.)

    I think video games are attempting to blend these two into some new form of human expression, experience. But I don’t think we’re anywhere near that yet. Maybe at times it comes close. However this is a deep cave system, and we’re only at the entrance.

  15. Brennen: I agree, the term ‘narrative’ is problematic. ‘Drama’ and ‘sport’ are in some ways useful metaphors, but they’re also ambiguous and carry some awkward baggage. For example, their use to describe embedded and emergent narrative makes an assumption about our perspective on each: we watch drama and participate in sport. But people also watch sport and act in dramas, both on stage and in their real lives; to an actor who liked to watch footy, the terms could be reversed and still carry a large part of the intended meaning.

    I’m also resistant to characterising this substantially new form of media in terms of what has come before. Obviously there is continuity with the media, games and real experiences that existed before videogames, but they are their own thing. Talking about them with words designed for fundamentally dissimilar concepts tends to cause confusion, as well as lend certain assumptions about what is desirable in a game (see Heavy Rain as an example of a mediocre game that was designed to be a great movie), which we may not have desired if it weren’t for our preconceptions.

    At RKD, we’ve made a conscious choice to use the term ‘videogame’ instead of ‘video game’, as the latter suggests a medium defined by pre-existing concepts that may not always be relevant: ‘video’ and ‘game’. Few videogames are played on video screens any more, and many videogames are not games in the traditional sense: most of Will Wright’s work, for example, are not games in the traditional sense because they lack a win or lose condition, i.e. an end. (Wright himself describes them as ‘software toys’.) We think the compound word, ‘videogames’, is somewhat more suggestive of a distinct entity in its own right.

    ‘Narrative’ is still a problematic word, though. Through past use, it’s assumed to mean a static progression; however, it’s more flexible than ‘story’, which is why I avoid ‘story’ a lot of the time. (I prefer simpler words where possible!) ‘Expression’, ‘experience’ and ‘meaning’ all capture aspects of gameplay more accurately, but lack in other areas.

    As you say, it’s an ongoing journey of exploration. We can only plough ahead with the best tools we’re able to fashion and wait to see how it all shakes out in the end.

  16. Austin:

    I saw your comment at Kotaku and posted a reply; unfortunately, Kotaku’s comment moderating system doesn’t know me from a pack of biscuits, so it hasn’t come through!

    A few important points to note:

    There’s no judgement implied with this theory. It’s only a descriptive frame: an aid to a clearer understanding of how specific narrative elements are created. (One commenter at Kotaku assumed I was judging the death of Aeris as the greatest moment of game narrative, because I put it at the top of the triangle!)

    Pure procedural narrative is probably impossible. As I mentioned in another comment above, it’s significant that there is no example on the bottom-right corner of that first triangle graphic. The nature of computer code is that it is always created by a human, even if indirectly. However, the further away an outcome of a game simulation is from what the designer intended and predicted, the closer it is to an abstract theoretical point where it was created without human input at all. That’s why I use a glitch as an example, as it’s an unintended outcome of the code, and therefore further away from the programmer’s intentions than ‘working’ code.

    Not all my examples are bugs, though. In fact only the first graphic mentions a bug. The examples from GTAIV are all correctly functioning processes within the game. Again, these are only partial examples of procedural narrative, due to the initial input of the game designer, but their computed rather than pre-stored nature makes them more a creation of the computer than, say, a cut scene.

    I also want to emphasise that the purpose of the triangle is to define an area rather than three discrete points. To my mind, narrative elements don’t usually belong in one category or another, but rather exist as the fluid result of multiple interacting factors. Pure procedural narrative is the most abstract and probably impossible concept, but I’d argue that pure player narrative and pure created narrative are also impossible, as any game situation only exists as an interaction between a computer, a player and a game designer, even if one of the three is relatively passive.

  17. Fascinating article – exactly the kind of stuff I am looking for. I’m currently in my last semester of a degree in Creative Writing and about to (attempt to) create a mini-thesis on the relevance of games and new media as modern fiction, I was wondering if you have any other articles like this? Do you mind if I quote you (my mini-thesis will never be published I promise you) ? and finally do you have anything you would suggest for me to look at? Let me congratulate you again on a well written and intriguing article.

Leave a comment