Interactive Digital Storytelling: Towards a Hybrid Conceptual Approach

1 Introduction In this contribution, Interactive Digital Storytelling is viewed as a hybrid form of game design and cinematic storytelling for the understanding and making of future learning and entertainment applications. The paper shall present formal design models that provide a conceptual bridge between both traditional linear narrative techniques as well as agent-based emergent conversations with virtual characters. In summary, a theoretical classification of thinking models for authors and interactive experiences for users will be presented. The conceptual work is based upon practical experiments within several research projects on edutainment, which employ conversations with virtual characters to convey information and to entertain. By building several prototypes, two different approaches where explored to combine plot-based interactive storytelling with character-based emergent conversations. Visual impressions of the examples are shown in Fig. 1 and will be explained in more detail in the full paper. In both examples, several virtual animated characters converse digitally with each other and with a user who mainly types text with the keyboard, optionally complemented by choice functionality and special hardware interfaces. The resulting conversations differ in their direction of approaching a middle ground between predefined narrative presentations and emergent conversations with a user, by combining emerging chatterbot dialogues with a story structure. The user experiences a semi-autonomous behaviour of interacting agents. This paper is not about the difference between stories and games. The motivation is on the potential of both to offer structures for learning and entertainment. Instead of trying to draw a distinct line between them, conceptual models for authors have to be defined, who are responsible to flesh out a suitable design within a variability of forms. Design elements include aspects of drama and filmmaking, dialogue design, as well as game design and game tuning. The actual challenge for the design of learning applications with conversational agents is the necessity that authors have to take on responsibility concerning the intended outcome and effect. In fact, they have to balance the bias between a pre-structured storyline (and possibly a timeline) which they may have strictly defined, and the agency that users shall experience through the design of the author. However, there is no one-dimensional borderline between both. In the following, the paper presents a model with several levels which shall help to form a more differentiated picture. 2 Conceptual Models for Storytelling and Agency In Fig. 2, a traditional modus operandi is sketched at four abstract levels. The distinction between levels may vary from project to project. The four levels were found to be suitable for the addition of interaction at each of the levels to form a classification of genres. On the top level of highest story abstraction, the overall dramatic outline is sketched. For example, there may be a hero’s journey in 3 acts, or a Propp model. Further, authors break down the story into scenes which are handled at the next level. Each scene will be defined by a scene script. Within a scene, dialogues and interactions between actors are defined, and lead to stage directions. If producing for an animated film, these directions are strictly mapped onto virtual actors by a skilled animator, who defines the way the virtual actors move and behave. When storytelling gets interactive, the user can influence the storytelling. In fact, in games as well as in constructivist scenarios for learning, users need to experience agency within a story. However, there are different levels at which to affect the outcome. In Fig. 3, the first author model (compare Fig. 2) has been extrapolated according to the need of introducing agency at each level. Opposed to the author, a participant is modelled who now may contribute to each level. The first implication for the author is that it’s not enough to just model a database of descriptions, but to add rules and models, which control an autonomous behaviour at each level in reaction to the participants. Then, it is possible to think of gradations of granted agency versus authored determination. Within Fig. 3 this is indicated by the sliders between control and autonomy at each level. The levels rather represent conceptual stages for authoring than elements of software architecture, though there are parallels to architectures of a number of existing systems of game and story engines. Semi-autonomy occurs on the edge between predefined factual information and rules for each level. The more rules on one level, the more agency can be experienced by potentially affecting the respective level. For example: It is imaginable that participants only experience agency on the lowest level, as a feeling of presence in a scenario. In this case, everything is predefined, but avatars would still react with nonverbal cues to the visitor and recognize her, comparable to a virtual cursor that shows a live status. At the conversation level, participants can for example have agency in an entertaining and informative chatbot dialogue with the characters. They may even not be able to affect anything in the story logic, but participate at dialogue level with speech acts. Agency at scene level would mean to have real choices about the outcome of a scene, for example, the story of the game would have to change according to user’s actions. On the top level, players would influence the whole story of the application, if the "agency slider" would be at a 100% to the right. For example, a simulation such as "The Sims" (Electronic Arts) can be put into the classification here. For factual knowledge transfer in a didactic lesson situation, the highest level could stay predefined, while the lower levels allow for conversational interaction, however constrained. If authors only provide a rule base with little pre-scripted structuring, they achieve a conceptual model more like an exploration or gaming experience depending completely on the action of the player. While arranging the bias at each level to various slider positions, several abstract genres of Interactive Digital Storytelling can be rebuilt in the model, which helps to specify exactly what kind of user experience an application shall provide. It is a conceptual model that can be used to classify story-related games, and it particularly supports authors coming from linear media, stepping into interactive storytelling. 3 Further Work In the full paper, I will also tackle related work while comparing with other theoretical models between games and stories, including references of the taxonomies of C. Lindley, M. Leblanc, J. Klabbers, B. Laurel, C. Pearce, and traditional classifications such as of R. Caillois. I will give more examples how existing products of Interactive Storytelling fit into the classification, and raise the question if new genres have to be defined particularly for Interactive Digital Storytelling. Literature Caillois, Roger: Man, Play and Games. (orig.: Les Jeux Et Les Hommes 1958) University of Illinois Press, Reprint (2001) Hunicke, Robin; LeBlanc, Marc; Zubek, Robert: MDA: A Formal Approach to Game Design and Game Research. In: Workshop Proceedings: Challenges in Game AI. 19th National Conference on Artificial Intelligence AAAI (2004) Klabbers, Jan H.G.: The gaming landscape: A taxonomy for classifying games and simulations. In Copier & Raessens (Eds.) Level up: Digital Games Research Conference. Utrecht University (2003) Lindley, Craig: Narrative, Game Play and Alternative Time Structures for Virtual Environments. In: Proc. TIDSE 2004, Technologies for Interactive Digital Storytelling and Entertainment, Darmstadt, Springer LNCS vol. 3105, 2004 Pearce, Celia: Emergent authorship: The next interactive revolution. In: Computers & Graphics 26, p. 21-29 (2002) Spierling, U.: Conceptual Models for Interactive Digital Storytelling in Knowledge Media Applications. In: Proc. TIDSE 2004, Technologies for Interactive Digital Storytelling and Entertainment, Darmstadt, Springer LNCS vol. 3105, 200