The Uncharted Frontier:
New Nonlinear Approaches


CREDITS: Copyright © 1995 by Michael St. Hippolyte. All rights reserved. Reprinted with permission of the author.

NOTES: This article was originally written with an intended audience of adventure game developers (think Zelda). I have inserted some text to address areas where ARGs are different from these types of adventures. I welcome any commentary or thoughts on the message forum.

SOURCE: The Uncharted Frontier: New Nonlinear Approaches. Reprinted with permission at

Although the existing approaches described above are sufficient to create interactive stories of great beauty and power, there is no reason to suppose that they are the only approaches possible. And there is every reason to look for new approaches. Not all great novels are in one aisle of the bookstore. How do we get to the next aisle?

In analyzing the existing models of interactive storytelling, and in particular the maze model, we identified the features that make these models work. We can now integrate these same features into new narrative models, hopefully resulting in blueprints for the creation of entirely new genres of interactive stories, and at the very least illuminating directions for artistic exploration.

Surprisingly, we can find many candidates for such narrative models among stories in traditional media. The following ideas may serve as a starting point.

Interactive Theater And The Character Space Model

Several experimental theatrical productions over the years have sought to make the theater experience more interactive. One approach has been to produce a play on a number of separate "stages" (often adjoining rooms in a nontheatrical setting such as a house) simultaneously. Characters in the play move from stage to stage, as do the members of the audience as they follow the characters that interest them.

Editor's Note: Did anyone else think of The Beast and the Rally for the Anti-Robot Militia or any of the other rallies?

This suggests a possible model for an interactive story, which we may call the character space model. In this model, the viewer navigates among the characters in the story (the character space), and at any time sees only the part of the story in which the currently selected character plays a role. This is simply a variation on the maze model, with characters replacing rooms as the navigational targets.

Now we face a crucial question: what can the viewer actually do in character space? In a typical maze story, the viewer is able to move, look around, manipulate physical objects and perform various other physical tasks. We could simply give the viewer these same abilities but in vicarious form, through the character she has selected. However, this would run the risk of turning the selected character into a zombie, which would probably make the story less interesting. Perhaps a better approach would be to give the viewer the ability to manipulate the thoughts, motivations, and/or emotional state of the selected character. To avoid running into the familiar problems of nonlinearity, this ability would have to be quite limited and the number of distinct states would have to be kept small. This should still allow for a rich and interesting system, since research has shown that quite complex human behavior can be effectively modeled with a small number of variables.

Here are two potential advantages of the character space model over the maze model. First of all, by taking away the viewer's ability to move and act physically, we remove the viewer as a de facto character in the story. She may influence one character or another, but all of the characters and their possible behaviors are defined by the author, who can therefore control the chronology of the story much more tightly than in a maze story. Secondly, in many stories what goes on inside the heads of the characters is more important that what goes on outside; the character space of such stories is likely to be much more interesting than the physical space.

Editor's Note: Player/character interaction is certainly a defining characteristic of Alternate Reality Gaming (ARG). We've seen phone calls, email, chat, blogs, and probably other devices used to introduce characters, define events, and otherwise move the narrative forward. I believe that character development is the most important thing a game designer needs to consider when developing an ARG. No amount of plotting and puzzling substitutes for the interaction available and immersibility of a well-written character. To make the game believable, the characters have to be believable. They need to act consistently with their biases and backgrounds and, in some cases, even be indistinguishable from the players themselves.

Rashomon And The Multiple Reality Model

In the classic Kurosawa film Rashomon, various witnesses provide completely contradictory accounts of an event. The film does not indicate which recollection is correct; each account in turn is depicted equally realistically. The sense by the end of the film is that all we have seen is unreliable, and that no account is completely true (or completely false).

Rashoman is deliberately and pointedly inconsistent, but uses this tactic to make a coherent and powerful statement. This is a capability which could be of particular relevance to an interactive story, if it allows the story to abandon the pretense of a explicit, unifying reality in favor of competing, possibly inconsistent realities. To the extent that inconvenient consequences of the viewer's choices could be ignored, a story based on such a multiple reality model could give the viewer more freedom.

But the logical inconsistency found in Rashomon is not the only kind of multiple reality imaginable. A more subtle multiplication of reality occurs when the witnesses to an event view it in ways that are simply very different rather than contradictory. Rashomon's multiple realities are subjective but pretend to be objective (in order to convince the magistrate); dropping this pretense would allow each subjective reality to be judged and appreciated on its own terms.

How would the multiple reality model work in practice? We could implement it as a variation (once again) of the maze model, with the navigation occurring between realities as opposed to physical locations. These realities could be associated with characters, in which case it would work much like the character space model. Or these realities could be of a different sort: different levels of meaning, different planes of existence, etc. Moving from reality to reality may be intrinsically interesting enough that the viewer would no further interactive ability; in any case, even simple actions by the viewer could be multiplied in their significance by the number of separate realities in which they occur. The viewer's actions could thus be highly constrained without detracting from the viewer's experience. This in turn makes it much easier for the author to avoid the pitfalls of nonlinearity.

There is, however, a particular narrative challenge inherent in the use of the multiple reality model. For the story to be satisfying, a unifying force must be found to tie the pieces of the story together, the way the criminal investigation into the contradictory accounts does in Rashomon. The author must construct the multiple realities so that they interact with each other in some way. If done effectively, this interaction will do more than just hold the story together; it will most likely serve as the vehicle for the central message of the story.

Overseas And The Intersecting Stories Model

The 1991 film Overseas, directed by Brigitte Rouan, depicts the lives of three French sisters in Algeria during that country's war of independence. Each sister's story is considered in turn; as the sisters' lives intersect, we see certain events depicted more than once, each time from the point of view of a different sister.

The result is remarkable. The events in which the stories intersect become reference points, and like in Rashomon, each time we witness the event the point of view and the significance of the event are completely different. But unlike Rashomon, most of each story takes place separately from the others, and where the stories do intersect, the factual accounts are identical. What changes is the viewer's experience: details that are mundane and trivial in one sister's account take on great significance in the next, while previously important matters are reduced to insignificance. On a more subtle level, the different motivations and emotional lives of the sisters give each story its own tone; as the story moves through an intersection point, this tone illuminates particular feelings and reactions that are invisible in the other stories. The cumulative effect is to give greater depth to all the characters in the film, even the minor ones, as each successive story unfolds.

Another achievement of the narrative structure of this film is its handling of the interdependent chronologies of the three stories. The chronologies intersect frequently, and generally at significant points in one or more of the stories. Nonetheless, what we see in the earlier stories does not spoil the suspense of the later stories, and what we see in the later stories never seems repetitive even though we may have seen it before. This is due in part to the fact that the three stories have distinctly different dramatic structures: the mix of drama, romance, comedy and adventure is unique in each story. Yet it all hangs together, because the structure of each story is well suited for its main character. The resulting narrative whole is a reflection of the sisters themselves: each sister is completely different, and somewhat of a mystery to the others, but their connection to each other is convincingly real, deep, and powerful.

The narrative structure of Overseas is uniquely suited to interactive storytelling. In an interactive version, which could be called the intersecting stories model, the stories themselves represent the viewer's navigational space, and the intersecting events constitute the navigable links. This approaches a pure narrative exploration; in this regard the intersecting stories model relates naturally to the medium of hypertext, and should indeed work well as a model for hypertext stories. A different strategy would be to apply another layer of metaphor to make the narrative exploration appear to be something more immediate and real. For example, the intersection points could be defined as physical locations, and moving from story to story could be accomplished by physically traversing a location.

The biggest difficulty with the intersecting stories model is the problem of chronology. A few simple rules may be needed; for instance, when the viewer moves to a new story, he could be automatically positioned at earliest unseen event in that story. Of course, these rules could vary with the requirements of the story, and in some cases may be altogether unnecessary. Much depends on the scope of interaction available to the viewer within the confines of each story. A clever approach would be to construct the stories such that their interdependencies are reduced to seemingly unimportant details (as is often the case in Overseas itself), in which case putting them off limits may not seem burdensome to the viewer.