Shadowcaster: A Blueprint for a Different Narrative Game

This article is part of the Shadowcaster Project


By reinvestigating the nature of games and narratives this article attempts to bridge the gap between the two in new ways. Games are regarded as ludic and expressive simulations. Stories, and popular fiction in particular, are regarded as means of giving shape and access to fictional universes. Narrative games are ludic interfaces to similar fictional universes. Drawing inspiration from pen-and-paper role-playing games a few strategies of interactive storytelling are investigated. These games have 30 years of experience with strategies that manage to balance player freedom with more or less defined narrative goals. All this is combined in the description of a new type of narrative game. A game where the ludic resolution of short narrative episodes is the main means of manipulation of the world-simulation.


The challenge for game designers who want to create rich, open game worlds and tell interesting stories at the same time, is to move beyond the constraints of unicursal corridors or multicursal hub structures while keeping the player's attention on a storyline. And it is no easy task.

- Espen Aarseth (2005: 11)

The issue of games and narratives has been the subject of many hot debates in the recently conceived academic discipline of Game Studies. Many, if not all, scholars of games will acknowledge the difficulty in combining the freedom of action promised by games with the authorial control and closure of the well crafted tale. It has been the project of the ludologists to radically separate games from narratives. They argue that the most successful games have little narrative aspirations and good examples of 'interactive narratives' make poor games at best. The narratologists on the other hand look for strategies that can overcome the incompatibility of authorial control and the player's freedom of action. Although their work is inspired, it resulted in little practical application. For now, games and narratives remain hostile towards each other. Those hybrid artefacts that manage to combine games with narratives often struggle to compete with pure games or pure narratives. The apparent incompatibility of games and stories is the biggest paradox of narrative gaming.

One of the problems with games and narratives is the narrow definition of narrative that puts at its centre a causal plot and closure in the first place. It is the type of narrative exemplified by traditional novels and feature-length films. In short, it is the traditional ideal of narrativity that grew with the rise of traditional novel and mediated by the linear media of the printed word and cinema. In contemporary society this type of narratives definitely serves as the most typical form of narrative, but it is not the only one: oral narratives, television narratives and the narratives of graphic novel all deviate from the traditional norm, but are not as primitive as some advocates of traditional narrativity would like us to believe. The strengths of these types of narrative lies not in its construction of an elaborate plot, but in the paradigmatic oppositions that underlie it.

In this article I will approach the problem of games and narratives from a new perspective: for I think the gap between games and narratives can be bridged if we take a view of narratives that deviates from the typical narratives of novels and cinema. At the same time this calls for a re-evaluation of the expressive potential of games. I will illustrate this new perspective on games and narratives with a fictional game called Shadowcaster. This game is fictional in that it does not exist, yet. This article is also part of the Shadowcaster project, an attempt to design and publish a small, independent, and narrative game.

Games: simulation and gameplay

Video games imply an enormous paradigm shift for our culture because the represent the first complex simulational media for the masses.

- Gonzalo Frasca (2003: 224).

Games have been discussed in great detail by an increasing number of scholars. From these discussions a few common concepts emerge that describe the activity of gaming and the characteristics of the game form. Here I like to focus on three aspects that I think are central aspects of gaming. These are simulation, gameplay (both addressed in this section) and representation (addressed in the next section).

Johan Huizinga described games in terms of its rules and rituals as early as 1938. He notes that games consists of activity that is separated from normal live. Games create their own order, in fact, they consists of nothing but order, and in that respect differ from real life (1938: 25). From a contemporary point of view we have come to understand these aspects of games as simulation, which constitutes a defining aspects of games (see for example Aarseth 1997: 141, Frasca 2003 and Atkins 2003). Simulation is a cost-effective design strategy to create consistent systems that allow for interesting emergent behaviour and gameplay (Smith 2001). The commercial and critical success of simulation games (or sand box games) such as SimCity, Civilization and The Sims indicates the strength of this formula. Game-designer Will Wright is currently taking simulation to its next level with the world generating game of Spore that was a big hit at the 2005 Games Developers Conference (see Diamante 2005).

But we do well to nuance the notion of simulation in games. As Chris Crawford states in The Art of Computer Game Design (1982: 9):

A simulation bears the same relationship to a game that a technical drawing bears to a painting. A game is not merely a small simulation lacking the degree of detail that a simulation possesses; a game deliberately suppresses detail to accentuate the broader message that the designer wishes to present. Where a simulation is detailed a game is stylised.

A similar point is made more recently by Steven Poole who stresses, contrary to the popular opinion, that games do not and, indeed, should not simulate reality. Despite many claims to realism, games simulate things that could not exists in the real world: "a car which grips the road like Superglue, which bounces uncrumpled off roadside barriers; a massive spacecraft with the manoeuvrability of a humble bee; a human being who can survive, bones intact, a three-hundred-foot fall into water" (Poole 2000: 64). The part of the fun of playing games is that it provides us with the illusion of racing a car that in reality would take years of training to control.

The relation between the simulation and the simulated world is not as important as one might think. From a design perspective the quality of the simulation itself is more important than any quality inherent to the simulated world. The game of Tetris with its mathematical precision is a perfect example of this. The game is fun because the simulation itself is enjoyable, not because of the depth of the world is simulates (which Tetris literally lacks). Simulation in games is as much about setting up rules that are fun to interact with as about trying to simulate something in reality, and a game always needs to compromise between these two goals. This inherent 'unrealism' of simulation of games is an aspect that these days seems to be forgotten all to often. Many games make good use of the number crunching capabilities of the computer in creating elaborate simulations that are so rich and detailed that they can actually serve as training modules for professional soldiers and pilots. This focus on realistic simulation tends to overshadow another important aspect of games: gameplay.

Games need to be played and the activity the player engages in should be fun. Gameplay (as elusive as that term might be) resides in the (rules of) interaction between the player and the simulated world. On its lowest level this means the pushing of buttons and the controlling of a joystick or mouse to make an avatar move or manipulate some game 'tokens'. On a higher level from these 'moves' emerge strategies and complex game 'expressions'. These strategies are still governed by rules, the simulation of a game typically favours a few strategies which are implicitly implemented by the rules of the simulation. Neither are all strategies immediately apparent: designers of games are often surprised by strategies developed by players that were never intended by the designers, but which are by necessity consistent with the game simulation. Strategies that emerge from the game play in this way might be difficult to control or design, and almost impossible to contain. With practice the best players can pull amazing feats: even at the highest difficulty level the whole game of Quake can be played in little over 12 minutes. This is a game which length is estimated to be at least 40 hours for most players at normal difficulty levels.

Games need to be fun, no matter how serious the business of gaming or how serious the practice of playing becomes, the initial appeal of all computer games should be that they are objects with which it is pleasurable to interact. The interaction should therefore be varied and challenging on the one hand but not to difficult to master on the other. Game designers go through great lengths to optimise the learning curve of a game. When a game is to difficult to learn players might give up, whereas games that are to easy also loose their appeal due to lack of challenge. The balance between pleasure and challenge is delicate but game designers have become experts at finding the right balance up to the point they should indeed start thinking of themselves as learning theorists, as James Paul Gee (2003) suggests. There are many aspects that influence the balance and the quality of the game rules. Among these I like to highlight the related concepts of coherence and smart configuration of relative simple rules.

Incoherence is one of the worst sins of game design. Steven Poole observes that incoherence ruins game play more easily than any disbelief in the reality of the game. He illustrates this with the some examples of Tomb Raider where a rocket launchers fail to even dent a wooden door. The unrealistic rocket-jumps of Quake on the other hand, are not very realistic, but they are coherent with the game universe of Quake and enjoyed by the players (Poole 2000: 64-66). Closely related to coherence is an idea of fairness: games in which the computer opponents resort to cheating are frowned upon. Computer opponents can easily cheat by making use of information and resources not available to human players. In some games the rules for computer players are altered. In the first instalment of Civilization, for example, the computer players are given a boost at higher levels because they need less resources to build the same units and city-improvements or to research the same technology as the player. Coherence is best achieved by keeping the rules fairly simple. The quest against special-case rules and exceptions that clutter bad games is one of the main drives behind Crawford's The Art of Computer Game Design (1983). Interesting game play can emerge from relatively simple game rules as long as these can interact in interesting ways. Many successful games are driven by simple rules which outcome is varied because of the complex ways these interact. Harvey Smith (2001) adds an economic argument to this: when the rules are kept simple the game-designer can spend more time on fine tuning the gameplay and creating more gaming content.

Incidentally, the family ties between computer games and board games is prominent in this regard. Quality board game also depends strongly on relative simple rules that govern the interaction with and between the various game tokens. Without the processing power of the computer the rules tend to be more abstract, but the games are not necessarily less sophisticated. The quintessential example of simple rules that underlie complex games is Go, which has enthralled players for several millennia. Another example can be found in the game of strategy and negotiation Diplomacy. In this game every army unit has a strength of one, each region on the map can accommodate one army and every army can take one of three or four action each turn. All very simple. The twist is that every player writes down their orders, and all these are executed at the same time. The success or failure depends on the co-ordination of actions between allied players, hence the name Diplomacy. The rules book of diplomacy consists for a large part of examples that illustrate the outcome of various combinations of actions, which helps the players to resolve the sometimes very complex situations that can emerge from the application of simple rules and moves.

Games: Representation

In a computer game, a space of possible representational events is typically enabled through a simulation. The simulation is a procedural representation, representing rules, not events.

- Rune Klevjer (2002: 200)

Unlike some critics like us to believe computer games are significant cultural artefacts. Like all works of art, computer games can represent, reflect and inspire. The use of games as vehicles for different types of propaganda (most notably America's Army and opposing games) is testimony to this representational power of games. What is more, in my opinion, computers are revolutionising the mediation of cultural content and computer games are among the vanguard of this development.

In many ways computer games are a new type of cultural artefacts: they are semiotic machines. Games are not texts, they are engines that produce texts. The aspects particular to games (simulation, game play and presentation) are not present, or not immediately present, in the text produced by a game. As any player can tell you, to watch a game or a game session is not the same as actual playing. Game-playing challenges contemporary ideas about authorship and authorial control of the text. Both the player and the computer contribute in the production of game text. They have become, to some extend, co-authors. This aspect of gaming and the ludic production of texts has been hailed by some. Greg Costikyan (1994), for example, praises games as a "democratic artform for a democratic age". Games challenge the notion of the central authority of the text, and this puts games in direct opposition with traditional narratives (see Frasca 2003: 233). At stake might be the quality of the text which traditionally has been attributed to the genius of the artistic author. According to this view no amateur player or machine can be expected to produce or partake in the production a digital masterpiece.

Saying that games are semiotic machines and not texts, is not the same as saying that games do not represent anything. Simulations are often disposed to behave in a certain way, and this leaves a lot of room for ideology, especially in the stylised simulations of games. For example, Ted Friedman (1999) exposes the disposition of SimCity as follows:

Of course, however much "freedom" computer game designers grant players, any simulation will be rooted in a set of baseline assumptions. SimCity has been criticized from both the left and right for its economic model. It assumes that low taxes will encourage growth while high taxes will hasten recessions. It discourages nuclear power, while rewarding investment in mass transit. And most fundamentally, it rests in the empiricist, technophilic fantasy that the complex dynamics of city development can be abstracted, quantified, simulated, and micro managed.

Likewise, in a later article Friedman (2002) states that in Civilization II:

there still remain baseline ideological assumptions which determine which strategies will win and which will lose. And underlying the entire structure of the game, of course, is the notion that global co-existence is a matter of winning or losing.

One of the most vocal advocates of simulation as an alternative mode of representation is Gonzalo Frasca. He repeatedly points out that our knowledge of "simulation rhetoric" is still limited (Frasca 2003: 225) and that it will take time before we reach the same level of proficiency with simulation as we have with traditional narratives (ibid: 233). Frasca identifies four levels in a simulation "that can be manipulated in order to convey ideology"; these levels are (ibid: 232):

  1. The level on which non-simulated (and thus traditionally represented) events or fragments are incorporated in the simulation. This level is perhaps best exemplified by the 'cutscene', a cinematic sequence that interrupts the game and all that it has come to stand for.
  2. The level of the rules of interaction: the actions taken by the player can directly represent a real action. The violent nature of most games is most prominent at this level, as the player's actions are often aggressive. But the lowering of taxes in the SimCity example above is also of this level.
  3. The level of long term strategies and goals, ultimately the strategies that result in winning the game. This level is perhaps the most interesting from a rhetoric point of view, especially because the player will have to make considerable effort to develop these strategies, while their ideology is not immediately apparent. The effectiveness of 'leftwing' environmental policies in SimCity is a good example of this. The ideology of this level and the previous one does not have to be the same. This is probably best illustrated by Frasca's own 'newsgame' September 12th. In this game the only direct action available to the player is violent and aggressive (throwing missiles at terrorists) but as an overall strategy it is very inefficient as it invariably leads to only more terrorist emerging.
  4. The last level is the meta-level on which players can change the simulation itself. Frequently game-developers make available tools that enable players to modify their games. The developers still control to some extend the nature of these changes, even though this control is limited and the 'modding' community resourceful. Interesting subversions of games circulate on the Internet, frequently changing games far beyond the developers original intent, and radically changing or challenging the game's original ideology, as is the case with the Velvet Strike mod for Counterstrike (see

To sum up, at the core of games lies the simulation of a gaming world with which the player interacts. The goal of this interaction might be to bring about a winning state, but this is not necessary; the activity itself and the external goals set by the player are often enough to captivate the player, as the success of the sandbox games illustrates. The simulation and the interaction offered by the game are not neutral activities: simulation and interaction in games often represent some real-life activity or ideology. This aspect of gaming should not be overlooked, because it is often a source of criticism (games and violence), and more importantly it is also a dimension in which most games have hitherto showed little development.

Now, lets turn our attention to narratives and the ways in which narratives and games might combine.

Narrative: Folk narratives, popular fiction and story-worlds

Storytellers in all media and all cultures are, at least partially, in the business of creating worlds.

- Scott McCloud (2000: 211)

We create worlds

- Origin slogan

On the outset narratives strongly differ from games. In its most typical form that of the literary novel a narrative has an elaborate plot which is propelled by believable characters and psychologically valid motivations, which still manages to surprise the reader with its 'novelty' and twists. The plot itself is structured mainly in time, the dimension expressed most directly by the sequence of words. Furthermore, narratives are invariably structured in the past, turning stories in histories. The reader vests in the author the responsibility for the quality of the plot, focusing all notions of quality, artistry and authority in this one person.

These notions of narrativity are ill-compatible with the characteristics of the game. As we have seen, in games players contribute to the story: their actions drive it forward. In this way the narrative game can no longer be in the past, and no single author can be responsible for the execution of a game sequence (cf Juul 2001). Also games, as simulations are not only structured in time but also in space: Civilization tells "the drama of a map changing over time" (Friedman 2002); "the spatiality of computer games marks a shift in importance from narrative to geography where players have experiences, which are not centrally narrative based or confined to narrative experiences" (Flynn 2003); "before we can talk about game narratives, then, we need to talk about game spaces" (Jenkins 2002). It is these differences that for a long time have fuelled the ludology versus narratology debate: for how can a linguistic theory possibly deal with games when it was developed alongside a narrative form that is so far removed from the game form? This objection is valid, and will probably remain so for a while. But the danger is to mistake one particular form of narrative (the novel) for all types of narrative, and all to often both ludologists and narratologists have fallen into this trap. Especially when those other types of storytelling, those that draw on popular and oral traditions, are so much better compatible with games.

If we compare stories that from oral traditions such as fairytales and myths to the stories commonly found in the literary novel a few differences come to fore. 1) The former type of stories are much shorter in length and usually they have very simple, often chronological plots. Often this makes them appear inferior to the so much more elaborate plots of the novel. 2) Popular stories and myths are often metaphorical, the characters are arch-typical and not very realistic; they are often made to stand for certain things that tie in with the proverbial moral of the story. This moral seems to affect the spatial structure as much as the characters themselves, as space is often used in the same metaphorical way. What drives the story, then, is not so much its plot but the relations and interaction between the concepts on the metaphorical map that underlie these plots. And finally, 3) fairytales tend to be very repetitive. Many use the same arch-typical figures and, as the well-know studies of Vladimir Propp (1968) and Joseph Campbell (1949) have shown, many share the same basic plot as well. Because the plot of these stories are more or less the same, it is more interesting to look ate the variance in the particular resolution of the interplay of the characters and concepts. This allows different fairytales or different tellings of the same tale to explore thoroughly the contradictions and oppositions in its metaphorical dimension.

Contemporary storytelling in popular cinema and television shares many characteristic with the fairy tale. Geoff King (2000) identifies the 'frontier myth' as the prototypical tale that is shared by most Hollywood action blockbusters. The frontier myth resolves around an individual hero who is contrasted with the anonymous powers of a technological, hedonistic and bureaucratic society which often is the hidden, antagonistic power of the story. The agents of this power are controlled by the technology they use, whereas the hero controls the technology she uses. The story takes place in a 'frontier' setting, originally this was the American West, but today the frontier can have many different guises: deep space, the below the ocean, the desert or any other wild lands not yet touched by civilisation. It is a hostile environment that has a purifying effect on those who enter, and only those that can be delivered from the clutches of technology will survive. The frontier myth is not as much an narrative template as it is an conceptual template. It sets up a metaphorical space that includes the concepts of technology, society, the individual and the frontier and in which many narrative developments can be mapped. The plots themselves are not important, rather these stories draw their expressive power and relevance from its portrayal of these concepts. The simplicity, repetitiveness and moral character of this arch-typical plot makes it resemble the classic folktale more than any contemporary literary novel.

Likewise, John Fiske identifies narrative theory derived from the folktale as an important source for the study of the long-running television narrative "with its simple, repeated structures" (1987: 148). He observes that in television the interplay of different concepts ("paradigmatic oppositions" is the phrase he uses) is more important for the narrative developments than any "syntagmatic chain of events" (ibid: 145). The result is that many television series never reach closure: the paradigmatic oppositions are never resolved and every episode starts with the same basic set-up. From a pragmatic position of the broadcasting company this is an advantage as it allows it the broadcast the series in almost any order, and it does not require the audience to see each and every episode. What changes occur are very gradual or have a long build-up. There is an interesting parallel between television series and collections of mythical stories, such as the Norse myths. The latter is a collection of short tales that on the one hand are short self-contained stories and on the other hand are part of a larger body of text. The order of the individual stories of the Norse myths is not really important, although some are clearly part of the beginning of the cycle while others are clearly part of the end of the cycle. There are a few long developing storylines that establish something of a chronology: the creation and following destruction of the world, and the increasing antagonism of Loki, the trickster god. The many reference and foreshadowings makes the sequence of middle stories less important than the element of the story-world they focus on and how these element interrelates with the other elements of the story-world. The whole cycle of the Norse Myth can be said to have unifying theme which binds the myths together and gives it general shape, but this shape is contained in the mythical world all the stories combined sketch out.

John Fiske also points out an important difference between the old folktales and contemporary television narratives. The former stem from homogeneous societies, whereas the latter are part of a modern heterogeneous society. This requires modern tales to be more open and multiple in interpretation (ibid: 148). The heterogeneous character of television narratives, the way it is open to many different, often contradicting interpretations is one of the main reasons of the popularity of the television. This is one of the central arguments of Fiske's book.

Changes in society are also a reason for Joseph Campbell to distinguish between a past era of storytelling when the 'monomyth' held relevance and a contemporary era where this needs to change under the pressure of modern, democratic and scientific society (1949: 388-390). In this regard King's frontier myth with its themes that address technology and the individual in an expanding society seem to fit well with the questions or at least some of the questions of this day and age.

These days also see a rise of fictional enterprises that resolve more around a fictional world than anything else. Star Wars and Lord of the Rings are the best examples of such enterprises. At the centre of the these lies a core story, but surrounding it is an increasing number of secondary stories and non-fictional texts. Fans contribute by producing fan fiction, while companies publish books about the story, setting, science or language. There are licensed games and theme-park rides. Series of comics and novels expand the original story or add completely new ones. It is a model that fits well with current business practises, economy and consumer behaviour, it should come as no surprise they are financially successful and very popular. In fact, these enterprises become so successful the fiction and its symbolism starts entering the real world: these days it is possible to have an wedding ceremony in Quenya or Sindarin, two of the Elfish languages devised by J.R.R. Tolkien. Most of these secondary 'texts' do not deal with the core narrative directly, rather they expand the fictional universe or are structured like a visit to the fictional world. They make the fiction seem more real by making it more complete and coherent and by giving the audience a chance to become part of it. More than then anything else folktales and popular fiction seem to facilitate a drive to enter a fictional world in which the reader is free to explore the symbolic play of her wishes, hopes and fears. Such fictional worlds lend themselves particularly well for simulation, and the interactivity offered by games make for a much more immersive experience. The sequence and causality of the plot are no longer the spine of the story, thus the need of an author who crafts such a spine diminishes. Instead the author (or development team) becomes responsible for the creation of a fictional world that is interesting, challenging and inspiring at the same time. The paradox between player freedom and narrative control starts to be resolved by reducing the latter to control over the particular disposition of the game simulation / fictional world. The paradox will be resolved totally when we start to realise that total freedom is not what the player wants, and we meet halfway between the two extremes of total authorial control and total freedom of action. As we will see below, most players are prepared to work with a game and actively structure their interaction with the game into a gaming narrative. Most of them are actively seeking a particular experience, are interested in play with a few, personally relevant symbols and choose the fictional worlds and type of game play of their leisure accordingly.

What I like to call 'narrative', then, has not so much to do with a recounting of a sequence of events structured as a plot. Rather I like to focus on the narrative function to experiment with, develop and communicate ideas in a symbolic and fictional universe. This is an activity that is often pleasurable and relevant at the same time. Most people have a narrative fallacy in a sense that they can recount and refer to real or imagined events in this way; people are very good in subjecting such events to a fictional world, and to learn from this. Narrative is a way of interfacing with our fictional worlds; narrative is a form of structured symbolic play; narrative is a particular signifying and interpretative mode. Conventionally stories are structured in time, but one cannot reduce narratives to those plays in which time is the most important structure. Narrative play is structured in all four dimensions, and everything these dimensions come to stand for. Stories are one way to share the narrative experience between people, but games can also be used to this end. Games designed for this purpose are what I call narrative games: storytelling engines that offer an interactive and often very immersive interface onto a fictional universe.

Narrative: Strategies of interactive storytelling

Adventures are of two kinds: underworld expeditions to the labyrinthine dungeons, or perilous treks in the wilderness. The former kind of game is typically the most varied, for it is played on a series of maze maps designed by the campaign referee, each map representing a successively lower level of passages and rooms deep beneath some weird castle.

- Gary Gygax (1974)

It is tempting to put games at the pinnacle of the development of narratives, offering a more interactive and personalised experience than other medium could allow. However, the newness of the technology does not guarantee a better way of storytelling. What it does offer is more and different ways of storytelling, enhancing the narrative palette with new tools and techniques that can be explored and exploited. These techniques are best put in the context of popular and folk narratives: they are an expansion of the arsenal that is used to create fictional universes that are open and meaningful to large and varied audiences. These techniques are also best used in conjunction with other, sometimes far older strategies of storytelling.

One interesting source of such strategies are pen-and-paper role-playing games. The best known example of these games is Dungeons & Dragons. Role-playing games have been around since 1974, making them younger then computer games by a few years. However, a varied and sophisticated set of narrative strategies has developed for and by role-players. In role-playing games these strategies were never hindered by technical restrictions that would reduce their stories in size, scope or vision; the main medium of role-playing games is the players' infinite imagination. Still, Dungeons & Dragons is not really known for its strong plots or dramatic developments. Originally the game was designed around dungeon adventures where the players had to explore a dungeon, kill the monsters and find the treasure. At the basis of these adventures is not a set of possible scenes but a map that outlines the dungeon, which has been prepared in advance or is taken from a commercial adventure module. The map provides details on the whereabouts of different monsters, secret doors, various pits and pendulums. In this way the dungeon map is the direct descendent of the battlefield of the miniature war-games Dungeons & Dragons evolved from: both set up and confine the dramatic potential of the adventuring session or oncoming battle. The existence of the prepared map contributes to the freedom by providing an easy and 'fair' method reference to the storyteller (or 'dungeon master'). It conveys the idea that the players can truly choose their own path and destiny; contributing to a sense of agency on the part of the player. On the other hand, the players cannot 'escape' the dungeon. There is usually no reason for the storyteller to prepare anything outside the dungeon. The map allows her to focus on the actions of the players within its confines, and this makes it possible to prepare the game and keep the number of possible courses of action manageable. Players are unlikely to try to go beyond the limits of the dungeon, because, after all, the whole point of playing Dungeons & Dragons is to explore the dungeon, slay the monsters and obtain the treasure. Most stories in Dungeons & Dragons work well as long as the players agree to try to stick to the role of the bawdy heroes the game casts them in.

Role-playing games have evolved from their 'dungeon-crawling' beginnings, but still maps are the backbone of many ongoing stories. The map is a way of simulating a world; designing a map sets up a web of possibilities for the players to explore. The old dungeon adventures are crude and primitive compared to the worlds and settings created for later games. These elaborate settings define the narrative disposition of the game by setting up an intricate simulation rife with dramatic potential. They have become story-worlds, even in those instances where a political or psychological 'map' forms its most defining structure. Such 'maps' appear in many adventures and modules of Vampire, The Masquerade. They often visually depict the intricate web of relations among the story's main characters. Figure 1 shows an example of one such 'maps'.

Figure 1: A relation map adapted from Berlin By Night, a sourcebook for Vampire, The Masquerade (p. 75).

A similar strategy is used by successful and critically acclaimed game-designer Will Wright who stated that:

As a game designer I try to envision an interesting landscape of possibilities to drop the player into and then design the constraints of the world to keep them there. Within this space the landscape of possibilities (and challenges) need to be interesting, varied, and plausible (imagine a well crafted botanical garden). It is within this defined space that the player will move, and hence define their own story arc. (Wright 2004: 13)

The downside of the story-world is that the player can easily become lost in its sheer size, especially when story-world is taken too literally. In most computer role-playing games that rely on huge maps the action quickly becomes repetitive. True, the advantage of computer role-playing games is that computers can store huge amounts of data. But that does not necessarily results in better games. How many dungeons should one visit to gain enough experience points to be able to deal with the next part of the story? For players interested in the hack-and-slash combat these games invariably offer, this is fine, but those who wish to progress the story can find this an arduous task. Story-worlds always run the risk the player will loose track of the narrative altogether, reason for Chris Crawford not to put too much hope in this structure (2003: 261-262). Still, when a story-world is used to guide and confine the possible directions of play and as long the players choose a particular game because she likes the type of play it has to offer, this strategy of interactive storytelling yields good results.

Narrative: Fractal Storytelling

Remember, gamewrights, the power and beauty of the art of gamemaking is that you and the player collaborate to create the final story.

- Orson Scott Card (quoted in Friedman 1999)

A more complex and perhaps more sophisticated strategy has been identified as fractal storytelling by Marie-Laure Ryan (2001) in her discussion of Neal Stephenson's science-fiction novel The Diamond Age. This novel features an interactive book designed to educate little girls by telling them stories and adopting the stories to reflect their real life. The book uses an extensive database of typical folktales and adopts these to fit the situation at hand. Nell, the main character and owner of the book, becomes an accomplished reader of this book; she is able to recognise these patterns and anticipates accordingly: "She had just entered the land of the oldest and most powerful of all the Faery Kings. The many castles on the mountains belonged to all of his Dukes and Earls, and she suspected that she would have to visit them all before she'd gotten what she'd come for" (Stephenson 1995: 308-309).

Narrative patterns in many genres are pretty easy to recognise, and the audience usually anticipates on these patterns. This anticipation forms the basis of suspense, and in a book or film correct anticipation can be a form of pleasure ("I knew this would happen"), in a game it can give the player a tactical advantage. We consume a surprising small number of stories while knowing nothing about them at all. We read the back cover of books before we buy them, we choose games because of the fun promised by its packaging. We plan a evening of television using descriptions in the newspapers, and we go to see a film because somebody told something good about it or because we read an interesting review. Film companies are aware of the importance of such information surrounding a film they actively try to direct such reviews. Jason Scott (2005) gives an interesting example of this with regards to Se7en: the producers of this film focussed on the thriller aspects of the film in the marketing campaign and information provided to the reviewers, while it carefully left out any connotations towards the buddy-cop theme that might be read from the same film. Likewise, most role-playing games are very clear about the promised type of play. Whether this is to enter a "grim world of perilous adventure" (Warhammer Fantasy Role-Play), "epic roleplaying in legendary Britain" (Pendragon) or "a Storytelling Game of Personal Horror" (Vampire: The Masquerade). Such dedicated systems work best when used for the type of story they are intended for. For example combat orientated play is not a very good strategy in Call of Chtulhu, a horror game in which monsters are very lethal indeed. On the other hand, true to its classic horror nature, the same game does offer detailed mechanics to deal with the player character's dwindling sanity.

The resolution for the narrative gaming paradox offered by the strategy of fractal storytelling is to work with these expectations and to confine the story accordingly. After all, when a princess is captured by a dragon we all know that the prince will rescue her. There is not much point in trying to create a game with several different endings to this age old challenge. It is better to direct energy in creating a game where this goal can be achieved in dramatically interesting and varied ways. Fractal storytelling is a way define the destination of a plot, although this does not necessarily fix the meaning, nor does it rule out all variance. The inevitable defeat of the dragon does not dictate how the hero will feel while doing this: he may remain indifferent, feel sympathy for the poor creature or maybe redeem himself in the act of killing. Each of these stories attach a different meaning to an apparently similar plot point.

A very interesting example of this structure is encountered in the Chinese film Hero. In Hero the same story is told again and again. The climax of the story is also always the same: a duel between the nameless hero and a character called Flying Snow. However, because the events leading up to this scene change a little with each telling the emotion that drives the scene changes from jealousy, to love and honour, each marked by a different dominant colour giving the scene a new poetic significance with each iteration. Hero more or less inverts the typical story tree solution commonly found in interactive narrative games. Where the typical story tree starts at one point and branches out from there to a number of alternative resolutions of the initial events, the different tellings in Hero converge towards a single point, which meaning and significance differ in each instance because the path towards it is different each time. When adopting a similar strategy, the interactive storyteller can direct more effort towards delivering an involving climax.

It is in this regard that Rune Klevjer identifies an interesting function of the cutscene in games: "it is a narrative of pre-telling, paving the way for the mimetic event, making it a part of a narrative act, which does not take place after, but before the event. The cutscene casts its meanings forward, strengthening the diegetic, rhetorical dimension of the event to come" (2002: 200). This is obviously not a function of every cutscene, but there are many examples of cutscenes that cast their shadows forwards, with the following game sequence being a more elaborate ramification of the same narrative structure. Such cutscenes can prepare us for the next sequence by displaying effective strategies or the preferred course of action. In this way cutscenes are used to great effect in Prince of Persia: Sands of Time where a few clues of the path towards the next save point. In the long run cutscenes can direct us towards the story's climax and resolution.

However promising the strategy of fractal storytelling seems, it is not without risk. If the path towards the goal is narrow, the players can be easily frustrated because they lack control over the destinies of their characters. In role-playing games this style of storytelling that leaves little or no room for the players to change its course is often referred to as 'railroading'. Players might feel as if they are stuck on a fixed track and their actions matter little to the final outcome of the story. It is a feeling the interactive storytelling must try to avoid, as it can quickly lead to frustration. Still many examples of successfully computerised interactive stories are heavily 'railroaded'. Maybe the best example is Half-Life, and not only because rails and trains feature so prominently in this game. In Half-Life the course of the action and the story is more or less linear. The player progresses through the levels of the game in sequence or fails and dies. There is no alternative. However, this feels hardly restricting as the course of action follows a well-known science-fiction, horror, survival template. The player needs to escape alien monsters only to discover that the government is not to be trusted, so he needs to escape them too. The narrative of Half-Life is delivered by positioning scenes that advance the storyline in the path of the player. The player is rewarded for paying attention to the story by better preparing her for the action at hand. These little scenes and the genre template in general function much like Klevjer's cutscenes, they tell the player what is expected and hint at things to come.

Bringing it all together: Shadowcaster

The object of this article is not only to outline some ideas about games and narratives, but also to present the blueprint for a narrative game that brings these ideas in practice. The game Shadowcaster is that game. Shadowcaster, is a pet project that grew from innumerable attempts to create a narrative game that I could enjoy. Because I wanted to play it myself it requires the ability to generate both levels and stories or quests. Shadowcaster is now also a collaboration with Jasper Rijnbout, who has recently started to work on the graphic and physics engine for the game.

At the heart of Shadowcaster, like in any good game, is a simulation. It is the simulation of the fantasy city of Sherrington. The town is officially ruled by a prince but the power is really distributed over six factions which uphold different ideals and ideas about running the city:

  1. The Nobility is trying to maintain the status quo. They are conservative and powerful.
  2. The Merchants are trying to dominate Sherrington's economy, they bring wealth to those who deserve, but have little sympathy for those that straggle.
  3. The Conclave of wizards govern all matters arcane. They quest for knowledge and magic skill but are blind for the needs of the mundane.
  4. The Guild of thieves have little scruples or respect for authority, but they look after each other and their neighbours.
  5. The Cult employ dark magic and are hedonistic, but they deal out the blessings and curses of their dark gods to all in equal measure.
  6. The Clan are secretive, destructive and beyond despair, but also a refuge to those who have lost everything else.

These six factions have a power-base among the three different castes of Sherrington: the common people, the magicians and the wealthy elite. All have need for Shadowcasters three main resources: food, magic, and money. The player character is a Shadowcaster, trained in the art of the swordmanship, stealth and shadow magic. This positions her right in the middle of all the factions, making her of interest to all factions and giving the player the opportunity to interact with them all. The quests the character undertakes and her relative success or failure in resolving these affect the power balance, and this is in turn reflected by the state of the city. When the Nobility grows in power, their palaces will glow with splendour but at the same time the common people might suffer. When the Clan gains influence, debris will clutter the streets. Rats and diseases will run amok among the populace. The Guild might gain control over certain areas but they might deal ruthlessly and harshly with those who question their ways.

There is no real goal for Shadowcaster, although some states can be considered end states. When one of the factions control over the city is complete there will be little material for stories left. The player can align herself with one of the factions and manoeuvre herself into a position of power, this might satisfy player defined goals. The player should be aware that her actions affect the power balance and should be encouraged to play around. The growth of the player is not reflected in the increase of statistics, such as strength, as is common in many role-playing games (both the computer and pen-and-paper variants). Instead the player and her character should grow into a role in the city of Sherrington. The player can still gain special powers or items that bestow special powers on her, but these should be role-oriented or provide new and interesting ways the character can interact with her simulated surroundings.

Shadowcaster will appear as side scrolling platform game, with an overview map that connects various game locations. Some areas will be action levels where the player will try to achieve various goals, while others represent other locations such as the city hall, inns, guildhouses, markets and private homes, where the player can interact with other characters. This presentation is chosen for several reasons:

  1. Shadowcaster is a two man effort and a side-scrolling platform game is within the scope of what we could reasonably create.
  2. A lack of a true 3D world or 2D isometric world makes it easier to insert locations and levels.
  3. 2D levels are easier to create on the fly.
  4. We like the platform gameplay.

There will be two modes for play in Shadowcaster, the first mode is the explorative mode in which the character can freely move around the city and interact with its various inhabitants. The other mode, of most interest to us here, is the narrative mode, in which the player selects and plays a new episode. An episode consists of a narrative development with the main character at its centre and which resolution will affect the power balance in the city. Each episode is condensed and should be played in about two hours or even less. We aim to let the length of each episode coincide with the length of a game-session. Ideally, a session of play might consists of multiple episodes played but no episode should span multiple sessions. The player develops her character over a series of episodes that are linked because they all draw upon the same settings and characters. Shadowcaster differs from other narrative games because there is no main story-line that spans multiple sessions, and there are no side-quests. Each session / episode presents her with a narrative situation which she will resolve one way or another. There are no save-games, there is no going back.

A simple quest episode might start with the player character finding herself in an inn (there is no need for continuity of space and time between episodes, when necessary the game cuts directly into the action). There she is asked by the friendly innkeeper to retrieve a certain item which he lost in the sewers. At this point the character might set out and try to retrieve the item in a series of typical platform game action levels. If she dies these levels restart and she can try again, until the time reserved for the episodes runs out. If she retrieves the item before that time the mission is a success and she gains the respect of the innkeeper, otherwise the mission is a failure and the player looses status in the innkeepers eye. The innkeeper might be a member of one of the six factions and helping him strengthens his faction, while failure to retrieve the item might weaken it.

However, things do not need to be as simple. Before the player character leaves for the sewers she might be addressed by somebody else, who claims that the innkeeper is really in league with the Clan. This new character claims to be a member of a faction who opposes the Clan and would gladly use the lost item against them should the player character bring it back to him. Now the player is offered a choice, but can she be sure the second character is who he says he is? There are many other examples of how various episodes might be constructed. Early episodes should give the player a chance to familiarise herself with the factions and inhabitants of Sherrington, while the later episodes have an increasing effect on the city's power balance (can she uncover the Guild spy who infiltrated the nobility? can she make a member of the Cult repent and make him join the ranks of the Covenant?).

Episodes are based on generic templates, the database of which should be easily expandable. The template for the example quest described above might be titled as 'A simple mission', and dictates that the player character should be given a straight forward retrieval mission by a character she knows, and be given a counter mission by a member of an opposing faction. There are two roles: a 'sender' and an 'anti-sender' (to borrow nominators from the semiotic actant scheme) and an action level that contains a goal. When the episode is initialised the program selects two characters who best fits the two roles and generates a sewer level to hold the lost object according to the experience of the player character and difficulty level set by the player. The program should be able to generate new characters if it cannot find suitable characters to fit a particular role. This requires the program to keeps track of the history of Sherrington, and log the different characters' interactions.

Needless to say, interaction between characters (whether controlled by the player or not) plays an important role in Shadowcaster. To model this character interaction and history we will adopt a model outlined by Chris Crawford (2005). A bigger challenge is how to incorporate the interaction in the actual gameplay. In narrative games dialogue is conventionally dealt with presenting the player with a handful of replies to whatever a game character says. This form of dialogue that resembles the ineffective and problematic branching tree strategy for interactive storytelling, inspired more by the constraints of the input devices of game consoles than anything else (see Poole 2001: 121). For a central aspect of the game it makes for very poor gameplay, and this is probably a worse crime than being also very difficult to generate on the fly. Instead dialogue in Shadowcaster will need to be a stylised simulation of conversation with some interesting gameplay. Although we are not yet definite on how to implement this, dialogue is likely to feature some realtime activity on the part of the player, in which timing and subject selection will affect the flow of the exchange. The text itself will be largely symbolic, not unlike the gibberish and symbols the characters of The Sims use for language. Of course there will be need for explicit directions and propositions from time to time but these will be the result of a well-played dialogue.


Interactive storytelling is a Holy Grail that has attracted the attentions of many knights, but remains out of our reach.

- Chris Crawford (2003: 272)

At the moment of writing Shadowcaster is in its initial design stages, far from being a realised project. The aim of this paper is to outline some alternative ideas, not to claim the Holy Grail of interactive storytelling for ourselves. Shadowcaster is attempt to embrace those aspects that make both games and stories interesting. These aspects are found in the fictional worlds they both create (by simulation or representation), and the pleasure of interacting with those worlds. In a narrative game the narration should be integrated with the simulation and gameplay. Shadowcaster attempts this by making the outcome of short adventures affect the power-balance in its fictional world, and by promoting character interaction to one of the game's core gameplay elements. It achieves this by pre-scripting these effects, but to have the player's action and choices directly affect the simulation. There is no special set of missions pre-designed for when one faction gains dominance, nor are there special episodes by which factions gain dominance.

There is a lot of work still to do, the actual gameplay remains to be designed, and we can only begin to form ideas of the proper strategies to implement the episode templates. This article, then, is a first in a series of articles covering Shadowcaster's design process. For how much fun it is to design a game, Shadowcaster was conceived as an academic experiment; an attempt to chart some new waters for the wonderful medium that is the computer game.