A Narrative Theory of Games Espen Aarseth Center for Computer Games Research IT University of Copenhagen 2300 Copenhagen, Denmark +45 7218 5045
[email protected] ABSTRACT This paper presents a narrative theory of games, building on standard narratology, as a solution to the conundrum that has haunted computer game studies from the start: How to approach software that combines games and stories?
Keywords Storygames, narratology, ludonarrative model
1. INTRODUCTION Computer games generate many questions and challenges for narrative theory. The very first humanist article on computer games, “Interactive Fiction” by Anthony Niesz and Norman Holland (1984: 125) talks about games as an “enigma […] to literary theory.” Are games a type of narrative? If not, do they contain narratives? Is narratology useful for the study of games? Should the definition of narrative be modified or expanded to incorporate games? Can the study of games yield insights useful for narratology? Perhaps we instead need a parallel paradigm, a “ludology” (Frasca 1999) to understand games the way narratology is used to understand narrative? Regardless of the answers to these questions, the fact remains that computer games have emerged as a dominant cultural form in the sense that it influences other forms such as cinema, TV, literature, theatre, painting and music. Therefore, the study of these other forms ignores the rapidly evolving field of games at their own peril. But how should such theoretical explorations be carried out? When changing focus from one empirical field to another, it becomes crucial to examine the examination process and the tools used as well as the object of the examination. Do theoretical concepts such as “story”, “fiction”, “character,” “narration” or “rhetoric” remain meaningful when transposed to a new field, or do they turn into empty, misleading catachreses, blinding us to the empirical differences and effectively puncturing our chances of producing theoretical innovation? Critical selfreflection is a hallmark of scholarship, and a necessary virtue when we examine a phenomenon with critical tools developed for another type of phenomenon altogether. In other words, when we study games through the lens of narrative theory, the lens itself must be critically examined as well. When we apply the perspectives and models from one form onto another, our ability to assess the incongruities as well as the similarities between the two forms becomes critical. Man is a Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. FDG '12, May 29-June 1, 2012 Raleigh, NC, USA. Copyright (c) 2012 ACM 978-1-4503-1333-9/12/05... $10.00.
pattern-finding animal. It is extremely easy to find parallels, precursors, and points of overlap, and thus seduce oneself to conclude that A is a form of B. The responsible theorist, therefore, should take the opposite position as their null-hypothesis: A is not a form of B unless proven otherwise. Consequently, in the context of games and narratives, one must be careful not to assume the task of proving that games are narrative forms, but to look for evidence and counter-evidence with equal zeal. To further complicate matters, “games”, as Wittgenstein pointed out in Philosophical Investigations Investigations (1953: §66), is not a category that it is possible to define formally. So how can we know that the phenomena we call computer games are even games in the first place? Successful definitions of narratives has a very long history and appear to be easier to come by, so at least there is some fairly firm theoretical ground to stand on: Narrative theory. As the study of computer games gelled into an emerging institutional practice around 2001, the question of whether games are a form of stories has become a prominent issue among the practitioners, a touchstone used to si gnal one’s familia rity with the new scholarly field. The last ten years have seen a number of comments on the so-called “ludology vs. narratology” debate, but, ironically, very few have actually engaged the question of the relation between games and stories through a properly narratological analysis, using the basic concepts and models of modern narrative theory. Instead, this “debate” has been carried out on a meta-level, through comments on, and characterizations of, the debate itself rather than by direct engagement with it. This metadebate can be seen as a symptom of the birth pangs of a new academic field. Too often the positions taken have been unnuanced, untenable, and therefore unproductive: “Games are always stories” (Murray 2004, p. 2). “[T]he computer game is simply not a narrative medium” (Juul 1999, p 1). 1 Tragically, in the field of game studies the term “narratology” has changed meaning and does not refer to the academic discipline of narrative theory, but to a more or less mythical position taken by an imagined group of people who are seen to believe that games are stories. It is high time, and hopefully not too late, to reinstate the original meaning and function of narratology, and ground the debate in narratological terminology and theory. Not to prove that all games are narrative (they are not) but to show that there is much to gain from a rigorous application of narratology to game studies. What has so far been lacking is a detailed, robust understanding of the various ways computer software have been used to combine elements from narratives and games into a number of quite different ludo-narratological constructs. What also needs to 1
A position from which Juul later wisely retreated (cf Juul 2001).
be realized is that story-game amalgams primarily is entertainment software, works that contain many forms of media content and because of their computer-based, Turing-complete existence can emulate any kind of semiotic genre, including, of course, traditional stories. Calling works like Max Payne or FallOut 3 games or stories is a metonymical shorthand usage of the terms that confuses and obscures the composite makeup of these creations.
2. Ludology vs. Narrativism The debate of whether games are stories has suffered from a lack of rigorous, theoretically grounded reflection and also from a basic confusion between normati ve and desc riptive approac hes. In reality this is not one, one, but two debates two debates conflated: one is the designoriented discussion of the potential and failings of game-based narratives, and another is the discussion of whether games can be said to be stories. The former is normative and partly speculative, partly critical, and the latter is descriptive and theoretical. These two debates have a partial overlap of participants, usually and unfortunately identified by the terms “ludologists” and “narratologists”2 but one debate concerns the viability of a semi-utopian hybrid game-story genre, and the other is concerned with the seemingly conflicting definitions of games and (in particular) narratives. The “ludologist” position was not, as has been claimed, “to see the focus shift onto the mechanics of game play” (Jenkins 2001) but to emphasize the crucial importance of combining the mechanical and the semiotic aspects and to caution against and criticize the uncritical and unqualified application of terms such as “narrative” and “story” to games. In other words, the ludologists’ critique was a reaction to sloppy scholarship (in which key terms are not defined), one-sided focus and poor theorizing, and not a ban against the application of narrative theory to games as such (an act they all had committed themselves): I wish to challenge the recurrent practice of applying the theories of literary criticism to a new empirical field, seemingly without any critical reassessment of the terms and concepts involved. (Aarseth 1997: 14)
That this challenge has been mistaken for a ban on the use of narrative theory in game studies is nothing less than amazing, and perhaps goes to show that humanist academics are often less astute readers, scholars and interpreters than their training gives them occasion to presume. It could also be suspected that anyone who echoes Jenkins’ misleading nomenclature of “ludologists” vs “narratologists” simply has not read the literature itself. Any attempt to clarify the relations between games and narratives will probably end up addressing both issues, but here the second, theoretical issue will be given priority. The question of whether games can succeed as a narrative medium does hinge on the question of whether games can be considered narrative at all, but
2
A terminology Henry Jenkins (2001; 2004) introduces to label the two sides in the debate. This was unfortunate, because it obscured the fact that all the so-called “ludologists” were trained in narratology and used narratology in their studies of games. In his influential article, Jenkins made it sound like the “ludologists” (Aarseth 1997, Frasca 1998, Juul 1999, Eskelinen 2001) were opposed to the application of narratology to games – “One gets rid of narrative as a framework for thinking about games only at one's own risk” – and not merely critical to the weak story-game hybrids at the time, and weak narratological applications. Jenkins’ article, published on his web site in 2001, ap pears to be the first tim e the word “narr atologist” was used as a label for one side in what he termed a potential “blood feud”.
answering it is not a job for theory, but rather for (future) criticism and, above all, creative innovation. As I have noted earlier (Aarseth 1997: 5), the difference between games and narratives is not clear-cut. However, games and stories seem to share a number of elements, namely a world, its agents, objects and events. It is crucial to note that these elements are also the cognitive building blocks of human reality, as well as of mediated representations of the same. It is thus fruitful to give priority to neither games nor stories, but rather to base the model in the primary reality that spawned both, and that they both are part of, in somewhat different ways. As pointed out above, it must be noted that “games” are not simply games, but complex software programs that can emulate any medium, including film, text/novel, graphic novel, and, for that matter, simulate board games and sports. We often commit the mistake of using the metonymic term “games” for software that in reality are integrated crossmedia packages, such as Max Payne (2001) which contains graphic novel pages and movie-like cutscenes (short animated movie clips that interrupt the gamePayne a story or a play), as well as ludic components. Is Max Payne game? Is it a hybrid? An amalgam? Whatever the answer, it seems clear that it is not purely a game, but a piece of software that does contain, among other things, a game.
3. The common denominators: What do games and stories have in common? There is not one, but many different techniques which have been applied more or less successfully to make “games” “tell stories,” and a ludo-narratological model of this design space must account for the ways in which “narrative games” differ from one another. There can be no single no single mode of narrativity in entertainment software, given the diversity of design solutions. To mention a few examples, MYST , Knights of The Old Republic ( KOTOR) KOTOR ) and Half-Life occupy Half-Life occupy very different positions in this design space, and a ludo-narratological model must reflect this diversity. My present approach is to see the ludo-narrative design space as f our independent, ontic dimens ions: WORLD , OBJECTS, AGENTS, and EVENTS. EVENTS. Every game (and every story) contains these four elements, but they configure them differently. Game worlds can typically be linear , multicursal , or open, open, and this has great effect on the game’s perceived narrative structure. Objects dynamic , user(including avatars and player vehicles) can be dynamic, created , or static, static , and again we see a span between the ludic (dynamic, simulated) and the narrative (static). Agents can be presented as rich, rich, deep and deep and round characters round characters (the (the narrative pole), or shallow, hollow bots (the ludic pole). The sequence of events can be open, open, selectable, selectable , or plotted , and the narratological notion of nuclei (kernels; events that define that particular story) and satellites (supplementary events that fill out the discourse) can be used to describe four different game types: 1. The linear game ( game ( Half-Life): Half-Life ): fixed kernels, flexible satellites. 2. The hypertext-like game ( Myst Myst , Dragon’s Lair ): ): Choice between kernels, fixed satellites. 3. The “creamy middle” quest game ( game ( KOTOR KOTOR,, Oblivion): Oblivion ): Choice between kernels, fl exible satellite s. 4. The non-narrative game game (Chess, The Sims): Sims): No Kernels, flexi ble discourse: just a game.
Five relevant ludo-narrative examples will be analyzed to illustrate the variable model: Oblivion, Oblivion , Façade, Façade , Fahrenheit (aka Indi(aka Indi-
go Prophe sy), sy), Half-Life Half-Life 2 and Knights and Knights of the Old Rep ublic. ublic. These analyses show that story-games display very different features along the four dimensions outlined above; in other words, there are many ways in which a game can be combined with a story, and so it does not make sense to look for one singular type of ludic story. Together with the metonymical problem, this diversity explains much of the confusion in the earlier debate and the previous lack of success in reaching a good, theoretical understanding of the narrative aspects of games. This model and analysis will demonstrate that narratology, properly applied, and combined with a broad sampling of different game types, can provide a fruitful and enlightening perspective on games and game design.
4. What is a Narrative? In the debate (insofar as it has taken place) it is sometimes argued that the standard notion of narrative is outdated and in need of expansion because it is poorly suited to describe games. Narratologist Marie-Laure Ryan suggests that “narratology must expand beyond its original territory” (2006: 98). Commentators such as Jenkins (2001) have suggested that narratives in games can be spatial, embedded and emergent. This conflation of any kind of diegetic or experienced situation with storytelling is what I have previously labeled narrativism (Aarseth 2004, see also Aarseth 1997:94 and Juul 2001). If an(y) interesting experience in a game is an “emergent narrative,” where does it end? And why limit this category to game-based situations? At some point it becomes hard to distinguish narratives from any other type of worldly experience, at which time (or long before) we might as well give up the discussion. Alternatively, we can consult existing narratology and see if the standard definitions of what a narrative is might not still be useful when we are examining games. Here is a model based th on classical 20 century narratology, which describes narrative and its constitutive elements:3
A hierachical model
(a pragmatic synthesis of many theories)
Narrative Story (what) Events
Discourse (how)
Existents Narration
Fabula
Sign chain
Things, places, Plot characters (Shuzet)
The chronological order of events of events
How the story is told (voice, focalization etc) The material stream of words, sounds or images
The arranged, unfolding order of events
3
I have, perhaps controversially, chosen to place sjuzet on the story- rather than on the discourse-side of narrative. The reason is that I subscribe to the idea that stories are dependent on the arrangement of events including selection of order. Oidipus Rex or A or A Study in Scarle t would not be the same stories if some other orders of selection were followed. However, the theoretical argument that follows is not dependent on this narratological position.
In relation to games, but also to naturally occurring phenomena in the real world, there are four elements in this model which can be said to exist across the categories: Events, things, places and characters. These are ordered by games, and narratives order them. This is the common ground, prescribed by narratology, which can be used to describe the relations between the two categories. In addition, the notion of what makes a story that particuthat particular story story is a crucial component that can also contribute to solving our question. I am thinking of the concepts of kernels and satellites (or constitutive and supplementary events, cf. Chatman 1978: 53-6). A kernel is what makes us recognize the story; take away the kernel and the story is no longer the same. If the wolf does not eat Red and her granny, the story cannot be recognized as Little as Little Red Riding Hood, Hood, so the eating is a kernel. D’Artagnan must befriend the three musketeers, or the story is not the one we find in Dumas’ novel. Satellites are what can be replaced or removed while still keeping the story recognizable, but which defines the discourse; replace the satellites and the discourse is changed. Red may stop in the wood to pick a flower, or she may not; this choice does not cause us to reject or accept a particular rendering of the fairy tale. Oidipus may eat dinner with his mom/wife, or he may not; either way we would not question the identity of the play. These two concepts, kernels and satellites, allow us to say something about the ways games can contain one or several potential stories.
5. World Gameworlds are physical or pseudo-physical (virtual) structures that are clearly delimited and which can be described by geometry or topology. They are different from so-called fictional worlds in that they, unlike fictional worlds, have a measurable, concrete extension that can be explored directly by an independent agent. Fictional worlds depend on the imagination, whereas game worlds have objective existence, even if they only exist via computing machinery. However, the world presented in a game is not necessarily a game world only. A game can contain two types of space, the ludic and the extra-ludic; extra-ludic ; the arena of gameplay, and the surrounding non-playable space. In certain games most of the space is extra-ludic, and the ludic space consists of narrow trajectories or corridors surrounded by static scenery. In other games, such as chess, the ludic space takes up the entire world. In others yet again, the players expand the ludic space by constructing more of it as part of the gameplay. In a previous paper (Aarseth 2005) I tried to to explain story-like games in terms of their quest structure, and used the shape of the ludic landscape as key to understanding how different game design strategies convey story material as part of the player’s discovery of the landscape through a quest journey. I listed three main types of landscape structure, a) the linear corridor ( Half Life) Life) the multicursal or hub-shaped labyrinth ( KOTOR, KOTOR , Far Cry 2), and the open world (Oblivion (Oblivion,, WoW ). ). If we add to this the one-room game (Façade) and separate hub from multicourse, we end up with five clearly different topological structures which have clear implications for the ease with which a particular story can be conveyed. To complicate matters a bit, we see that the five structures can be combined to form more complex patterns, such as a game with a linear beginning, opening up to an open quest world in the middle, and then closing in at the end to another Oblivion and Fallout Fallout linear corridor. Typical examples of this are Oblivion and 3. Also, games can consist of an open landscape littered with
smaller labyrinths (dungeons) such as the caves and instances we find in World of Warcraft .
6. Objects Objects in games can be categorized in terms of their malleability: a) static, non-interactable objects b) Static, usable objects c) Destructible (buildings in a RTS) d) Changeable (e.g. weapons in Resident Evil 4) 4 ) e) Creatable (E.g. armor in World of Warcraft ) f) Inventible (creatures in Spore, computers in Minecraft ). ). Of course, one and the same game can contain all of these categories, and most contain more than one type. They are important because they determine the degree of player agency in the game: a game which allows great player freedom in creating or modifying ob jects will at the same time not be able to afford strong narrative control.
7. Characters
can be influenced but not the satellites, would typically be a nonlinear story (a hypertext fiction) and not a game. Events can also be constricted temporarily, to let the story-elements be conveyed through traditional narration in an otherwise high-agency game. A typical example is Half-Life 2 where Gordon Freeman is sometimes immobilized and/or transported on rails through the landscape, or positioned in a closed room for the duration of some NPC dialogue.
9. The Variable Model After having described the variables possible in each of the four dimensions, it is now time to put them together in a single model:
Onti Ontic c leve level: l:
After universe, characters are the most important element in crossmedia productions. The characters found in games are sometimes imported from other media, and can be classified in terms of their depth/shallowness, and their malleability/potential for player control. The game characters can be categorized into three different kinds: a) “Bots” (short for robots), with no individual identity (e.g. the metrocops in Half-Life in Half-Life 2); b) Shallow characters (names and individual appearance, but little personality) and c) Deep characters (Trip and Grace from Façade from Façade,, Lucas Kane from Fahrenheit ). ). A clear parallel to the latter two categories can be found in E. M. Forster’s classic Aspects of the Novel (1927), where he makes the distinction between flat characters (who basically stay the same no matter what happens to them, and round ones (who change and develop as the story progresses). As with objects, the same game can contain a mix of these categories, and again the level of malleability determines the authorial affordance of the game. In addition, it can be claimed that the richness of character is an important authorial tool that characterizes the positive potential of authorship in games, where malleability and user control limit authorial affordances.
Narrative pole
Wor World ld
Obje Object cts s
Inaccessible
Noninteractable
Single room
Static, usable
Linear corridor
Modifiable
Multicursal labyrinth
Destructible
Agen Agents ts Deep, rich, round characters
Events fully plotted
Dynamic satelites/ playable story flat characters Dynamic kernels
Ludic pole
Hubshaped quest landsape
Creatable
Open landscape
Inventable
Bots, no individual identity
No kernels (pure game)
The four-dimensional model with game examples:
Onti Ontic c leve level: l:
Pure story (War and Peace)
Wor World ld
Obje Object cts s
Inaccessible
Noninteractable
Single room
Static, usable
Linear corridor
Modifiable
Multicursal labyrinth
Destructible
Hubshaped quest landsape
Creatable
Open landscape
Inventable
Agen Agents ts Deep, rich, round characters
Events fully plotted
Fahrenheit
8. Events
Half Life 2
Events can be categorized by the status and presence of kernels and satellites: a) fully plotted (pure story); b) dynamic satellites (playable story); c) dynamic kernels (multipath/quest games); and d) no kernels (pure game). A work in which the choice of kernels
Dynamic satelites/ playable story flat characters Dynamic kernels
KOTOR Oblivion Pure game (Minecraft)
Bots, no individual identity
No kernels (pure game)
Discourse/Influence Kernel Influence
Satellite influence
No influence Choose Alternatives Full influence
Not possible
Possible
A linear story story
A linear game game
(War & Peace)
(half-Life 2)
A nonlinear nonlinear story (hyperfiction)
N/A
creamy middle quest game
“
As seen here, the example profiles seem to suggest that the most important dimension for storytelling in games is that of agents/characters. This indicates that the most effective way of creating ludo-narrative content is to invest in character-creation, by making the cha racters rich, dee p and interesting. ”
(Oblivion, KOTOR)
just a game (Chess, Minecraft)
Story
e y y a m o r y o r t o t o G S s r r a a r n e a n e i n n e L i L i o n l
N
m e G a t s u e Q
m e G a
”
r e P u
“
Game
It should also be noted that the first two dimensions, World and Objects, are describing not so much narrative elements as player agency. Likewise, the two latter dimensions, Agents and Events, are describing not so much gameplay as author agency. Hence, we can make the observation that the only the latter two are narrative dimensions per se, while the first two are describing ontological aspects of the game world, rather than inherent “narrative” qualities of worlds and objects. There is nothing necessarily
narrative about topological variation in world structure, nor about the degree of flexibility of game objects. The removal of agency is not a measure of narrativity, even if it is compatible with story production. This also means that a linear world cannot be classified as “more narrative” than an open-field one, or that games with limited player-object agency by necessity are more narrative than others. It merely means that linear-world, static, object systems pose fewer challenges to ludo-narrative projects.
10. Conclusion The best way to test a model is to introduce new data and see if it fits the model. Two recent examples that seem particularly relevant are Heavy are Heavy Rai n and Dragon and Dragon A ge: Orig ins. ins. They are different in terms of game objects and world structure, similar in terms of characterization, with deep, interesting characters and where Dragon Age Age has a Dynamic Kernel approach, Heavy Rain Rain seems to use the Playable Story structure. Both put a lot of effort into characterization, as the model suggests they should. There is nothing in either example to suggest that the model needs revision. The “nature” of ludo-narrative works is complex and multiform, and yet there are a few basic ontological dimensions that can be used to describe this variance. The model presented here does not account for “content” aspects of ludo-narratives, such as emotions, themes, style etc. It could still be used to analyze and suggest design decisions, even if the main purpose is to be descriptive, not prescriptive. More refined versions could no doubt be produced to distinguish between variations that are not captured here. And, the possibility remains that one or more of these four dimensions can be replaced by others with greater explanatory power. The value of narrative theory in understanding and describing ludo-narrative variation is undeniable and profound. The once vigorous but theoretically weak discussion of 1) whether games are narratives and 2) whether narrative theory should be applied to this kind of entertainment software has long run out of steam, and should be put to rest. Instead, I hope to have shown that 1) “games,” as a metonymic label, is the wrong term for ludonarrative software, and that narrative theory, while necessary, is not sufficient to understand these new forms.
11. REFERENCES [1] Aarseth, Espen (1997): Cybertext: Perspectives on Ergodic Literature. Baltimore and London: Johns Hopkins UP. UP . [2] Aarseth, Espen (2004): “Genre Trouble: Narrativism and the Art of Simulation” in Pat Harrigan and Noah Wardrip-Fruin (eds.) First Pers on. on. MIT Press, 45-55. [3] Aarseth, Espen (2005): “From Hunt the Wumpus to EverQuest: Introduction to Quest Theory” in Proceedings, Entertainment Computing - ICEC 2005: 4th International Conference, Sanda, Japan, September 19-21, 2005. 2005. Lecture Notes in Comp uter Science, Springer Science, Springer 2005, pp 496-506. [4] Chatman, Seymour (1978): Story and Discourse: Narrative Structure in Fiction and Film. Film. Itacha and London: Cornell University Press. [5] Eskelinen, Markku (2001): “The Gaming Situation” in Game Studies (1,1) Aspects of t he Novel . Mariner [6] Forster, E. M. (1927[1956]): (1927[1956]): Aspects Books. [7] Frasca, Gonzalo (1999): “Ludology meets Narratology: Similitude and differences between (video)games and narrative.” http://www.ludology.org/articles/ludology.htm [8] Jenkins, Henry (2001/4): “Game Design as Narrative Architecture.” http://web.mit.edu/cms/People/henry3/games&narrative.html [9] Juul, Jesper (1999): A (1999): A Clash betwee n Game and Nar rative. Master’s Thesis. Available from Jesperjuul.net [10] Juul, Jesper (2001): “Games telling stories?” in Game Studies (1,1) ies (1,1) [11] Murray, Janet (2004): “From game-story to cyberdrama”, in Harrigan and Wardrip-Fruin: First Wardrip-Fruin: First Pers on. MIT on. MIT Press. [12] Niesz, Anthony J. and Norman N. Holland (1984): "Inte ractive Fiction" in Critical Inquiry Volume Inquiry Volume 11, Number 1, 110129. [13] Ryan, Marie Laure (2006): Avatars (2006): Avatars of Story. MinneapStory. Minneapolis/London: University of Minnesota press. [14] Wittgenstein, Ludwig (1953 [2001]). Philosophical [2001]). Philosophical Inve sti gations. gations . Oxford: Blackwell Publishing.
