From Tabletop RPG to Interactive Storytelling: Definition of a Story Manager for Videogames

From Tabletop RPG to Interactive Storytelling: Definition of a Story Manager for Videogames Guylain Delmas 1, Ronan Champagnat 2, and Michel Augeraud 2 1 IUT de Montreuil Université de Paris 8, 140 rue de la Nouvelle France 93100 Montreuil, France g.delmas@iut.univ-paris8.fr 2 La Rochelle University L3i, 17042 La Rochelle, France {ronan.champagnat,michel.augeraud}@univ-lr.fr Abstract. Adding narrative in computer game is complicated because it may restrict player interactivity. Our aim is to design a controller that dynamically built a plot, through the game execution, centred on player s actions. Tabletop Role-playing games manage to deal with this goal. This paper presents a study of role-playing games, their organization, and the models commonly used for narrative generation. It then deduces a proposition of components and data structures for interactive storytelling in videogames. A prototype of a social game has been developed as example. Keywords: Interactive storytelling, role-playing games, story manager, videogames. 1 Introduction For the last years, interactive storytelling has become a major issue in video games development. Players ask for both more elaborated scenarios and more freedom to act and affect game's story through their avatar. However, storytelling unfolding and player's interaction are commonly said as opposite [1]. The first relates to game designer's control of the game he has created as the second relates to player's control on the game he has bought. Research on interactive storytelling mainly focuses on this opposition. Works on scenario driven approach [2, 3, 4, 5] have defined approaches to design, validate and control a scenario through an interactive application. However this control is somehow presented as in contradiction with player's ability to influence the game's unfolding in a persistent way. On the opposite, emergent narrative theory [6, 7, 8] gives a complete freedom to the player, who can direct the game unfolding through his actions. But this approach currently fails to ensure consistency and an interesting unfolding of the game. The two main issues we have underlined are about the relation between interaction and narrative, and the ability of computer science to deal with it. The first point is related to interactive storytelling theory, and concerns the way interactivity should be taken into account in the storytelling. The second point leads to computer theory and the ability to implement algorithm of interactive storytelling. I.A. Iurgel, N. Zagalo, and P. Petta (Eds.): ICIDS 2009, LNCS 5915, pp. 121 126, 2009. Springer-Verlag Berlin Heidelberg 2009

122 G. Delmas, R. Champagnat, and M. Augeraud This paper begins with a presentation of the interactive narrative we target. It comes from tabletop role-playing game. A game master balances between the control of the narrative and the freedom for player. The next section presents the architecture developed. It is based on a multi-agent system that monitors and controls the game unfolding. This architecture has been implemented in section 4. 2 Interactive Storytelling in Tabletop Role-Playing Games Tabletop role-playing games (t-rpgs) arose with the emergence of social games in the end of the 19th century. The first example of commercialized t-rpg was Gary Gigax's Chainmail in 1971. Basics of t-rpgs are to let a set of players share the creation of their own story, where they interpret the main characters. T-RPGs mix dynamics from both society games (for the ludic aspect) and improvisational theatre. In the games' field, t-rpgs obtain results similar to the ones we aim at in the video games' field: an adaptive unfolding of the story, centred on player actions that is satisfactory for the player. T-RPGs share a common model. A set of players interprets characters (named player characters - PC) and a game master (sometimes referred as storyteller) acts as a story moderator and a referee. The game describes a specific environment (which can be completed and personalised by the game master) and a set of rules for game simulation. The game master is both a referee and a story director. On the one hand, he checks the player characters actions. He validates actions and their results according to game s rules. On the other hand, he is responsible for story s unfolding. He has to describe the environment and to interpret the set of non-player characters (NPC). As a consequence, he produces a frame for the story and adapts it to player s actions. At the beginning, the game master describes an initial situation to the players. It is a precise description of the story world according to PC s point of view. Each player tells the game master its character s actions. The latter computes the action (he applies the corresponding rules), and can add NPC s actions or specific actions. Finally he describes the resulting new situation. Then players can act in order to influence the story world and thus unfold the situation. Depending on the evolution of the situation, the game master chooses a story starter and presents it to the players. It is an action that destabilizes the situation and provides a problem the players will have to solve. As the players decide, or not, to try to solve the problem, and then to start the story, the game master has to provide an action and a problem that interest them. It is the players responsibility to tackle the issue, and also the actions they perform. However the game master can influence players decision by giving them specific information. The game master monitors the story unfolding, through player s actions, and estimates the most relevant conclusion. He aims to direct the story towards one of these conclusions. As a consequence he applies some actions (not depending on player s actions) to the current situation. He also checks that the story remains consistent, avoids deadlock or hazardous decisions, and manages the pace of the narrative tension.

From Tabletop RPG to Interactive Storytelling 123 The story derives from the interactions between the players and the game master. It can be a complex process and can be split into a sequence of task performed by the game master starting from a set of resources we will see in the following. 2.1 T-RPG Basics In order to produce interesting and open narratives, t-rpg applies specific interactions rules between players and game master. Rules can vary from one group to another. However, we can distinguish a set of constraints that allows a game master to create interesting and interactive narratives with the players. Players must not be constrained by the game: they have to be presented a set of actions (as inputs of the game) that allows them expressing themselves through their characters. These control elements must be persistent (the game master cannot inhibit them). The players should perceive the consequences of their actions. And finally these actions should have a real impact on game s unfolding (the narrative) and its final result (story s conclusion). The game has to propose a consistent environment: in order to favour player s immersion. Game s rules must seem sensible and cannot be changed during the course of the game. Game s unfolding has to satisfy to storytelling structure: it is not limited to a sequence of situations without linking the one with the others. It has to satisfy a structure that depends on the kind of storytelling defined by the designers. It is the game master s responsibility to maintain these constraints throughout the game. To this purpose game masters have to perform a set of specific tasks. 2.2 Tasks Performed by the Game Master Whereas most game masters determine their action through empirical and nonordered processes, we can discern four main stages in game master's decision process. Get and analyse information from the players. The game master identifies how the player action can be taken into account on the story unfolding and to identify what information they give on the player s behaviour and its character. The game master collects information on the player s knowledge. He maintains a profile of the player that will help him to make choices during the story unfolding. However, the main phase of the game master consists in synchronize the state of the game, identify the possible histories and to choose the sequence of event to perform. Finally, he has to determine the actions or information to give to the players in order to influence the story unfolding. The game master needs information concerning the game, the players and the narrative structure in order to perform these tasks. 2.3 What Should the Game Master Know? In order to realise this decision process, the game master can refer to three main sets of knowledge:

124 G. Delmas, R. Champagnat, and M. Augeraud The game: the game master both knows the game rules as well as the game world. He will anticipate the consequences of actions and solve the player interactions. The game world allows to determine the current situation, and to know which narrative resources the game master can use (characters, actions, etc.). A narrative structure: the game master uses a structure of story in order to have a consistent and valid unfolding of the story. It gives him a framework to organise the actions of the game and to know when he must put actions. Thus, the game master can improve the interactive story that is being unfolded by giving a strong structure and this will improve the player experience. Player s profile: the game master influences the unfolding according to its knowledge of the players (strength, weakness, etc.). He must, also, know their preferences in order to select actions that will fit to the desires of the player. This study of t-rpg provides us with information on what interactive storytelling can achieve, and how to obtain these results. [9] presents an interesting study of t- RPG as the base of narrative videogames, in particular by the reference to the game master as a controller for the computerized systems. We deduced from this theoretical study an architecture for unfolding of interactive storytelling in video games. 3 Architecture for Interactive Storytelling in Videogames We proposed in [10] a controller for interactive storytelling in videogames. We can sum up our approach of interactive storytelling in the following way: a dynamic construction of the story, centred on the player s actions in an environment of interactive game. So, this construction will not be based on a set of predefined paths, but on a control mechanism that analyzes the game unfolding and the player behaviour, and generates actions directing the storytelling towards one of its possible outcomes. We have defined a software architecture that allows the implementation of our approach of the interactive storytelling: the player controls the narrative, but not the game. The game defines an interactive environment in which the player can, by his actions and decisions produce a storytelling under the supervision of the controller. Controller s action is bounded by the game (defined by the game designer). Thus the controller cannot go against game s unfolding nor prevent player s actions. It is built on the top of four agents: Analysis, Profile, Story Management and Director. The analysis agent receives information from the game and sends them back to the system. It performs a first selection between useful and non-pertinent information. It sends to the story management agent all information about the state game evolution and to the profile agent all information coming from player actions. The Profile agent contains the knowledge database on the player. This agent merges all information on the player actions and derives statistics on its behaviour and its preferences (type of the most used action, game element that is generally used by the player, player performances, etc.). The Story Management Agent monitors the game unfolding and takes decision during the execution. He has to assure the monitoring of the game s state, the determination of the possible narratives, and the selection of the actions to perform the storytelling.

From Tabletop RPG to Interactive Storytelling 125 The Director agent aims at making a retransmission of the story management decision to the game. It transforms the story management decision into comprehensible directive by the game. In addition, the controller requires three views of the game: a game model, a narrative construction structure and a player profile. The game model gives all the possible stories and follows the game unfolding (determines the game s current state, as well as its possible evolutions). It is base on a computational model (Petri nets). The narrative construction structure gives a canonical model to the story (Aristotelian structure, The heroes journey of J. Campbell, etc.). The player profile is on statistical point of view but not on psychological, and contains information to identify player s preferences. 4 Example: The Schoolyard This example is a prototype to validate the architecture proposed previously. In this game, the player acts for a kid that comes recently in a new school. Lost in the middle of unknown pupils, he has to make new acquaintances and find friends. The virtual characters present in the game employ the same actions the ones towards the others. It is by taking their control that the system can influence the contents of the play. Characters take a prototype of schoolyard: the nerd (a boy fond of sciences, and presenting social difficulties), the sportsman (very popular and very friend with the cheerleader, but not appreciating the intellectual characters), the captain of the cheerleaders (very friendly with everyone, except the captain of class) and the captain of class (serious pupil, which gets along only with the nerd). Each character is characterised by a temper: a specific model gives the evolution of the temper and the different states, according to interactions with other characters. The temper is of four states shared by all the characters (avatar and non player characters): neutrality (default temper), sadness, angriness and happiness. The state temper evolution is threshold by interactions. Their friendly level or hostility level describes the interaction between characters (neutral, positive or negative). The player s profile is a short set of statistical information due to the fact that the proposed actions are limited (the number of positive/negative interactions, the number of interaction with a specific character, etc.). The game proposes two separate scenarios: the fist one is the relation network a player can create, and the second one the temper evolution of a character the player considers as a friend. This example is simple: the game has a basic narrative program: a contract phase and a qualifying phase. The knowledge phase can be divided into a fixed number of narrative sub-programs. 5 Conclusion We have presented a study on tabletop role-playing games and narrative management. From this study we have derived a set of components and data structures to control

126 G. Delmas, R. Champagnat, and M. Augeraud the interactive storytelling in videogames. These structures have been integrated into a control system for interactive storytelling in games. This proposition has been illustrated by the realisation of a social game that provides interactive experience through a limited interface. This prototype shows that our controller s architecture is able to determining a narrative during the game execution; check the controller ability to direct the narrative to a conclusion while satisfying narrative construction structure; and finally make sure that the controller decisions are in real-time. Although the example is basic, the controller has been designed in order to deal with larger games and with more complex data models, such as the story structure presented in [11]. In addition, the controller proposed in this paper has been originally designed for a narrative-oriented game but can be used to control a sandbox game as stated in [12]. Acknowledgments. This work has been funded (in part) by the European Commission under the grand agreement IRIS (FP7-ICT-231824). References 1. Juul, J.: A Clash Between Game and Narrative. In: Digital arts and culture conf., Bergen (1998) 2. Mateas, M.: Interactive Drama, Art, and Artificial Intelligence. Ph.D. Thesis. School of Computer Science. Carnegie Mellon University (2002) 3. Young, M., Riedle, M., Brandy, M., Martin, J., Saretto, J.: An Architecture for Integrating Plan-Based Behavior Generation with Interactive Game Environments. Journal of Game Development 1, 51 70 (2004) 4. Riedl, M.: Narrative Generation: Balancing Plot and Character. Ph.D. Thesis, Department of Computer Science. North Carolina State University (2004) 5. Champagnat, R., Prigent, A., Estraillier, P.: Scenario Building Based on Formal Methods and Adaptative Execution. In: Narasimhan, S., Teach, R. (eds.) Proceedings of the 36th International Simulation and Gaming Association, ISAGA (2005) 6. Aylett, R.: Narrative in Virtual Environnements Towards Emergent Narrative. In: Proceedings of the AAAI Fall Symposium on Narrative Intelligence, pp. 83 86. AAAI Press, Menlo Park (1999) 7. Szilas, N.: IDtension: a Narrative Engine for Interactive Drama. In: TIDSE 2003. LNCS, vol. 3105, pp. 183 203. Springer, Heidelberg (2003) 8. Cavazza, M., Charles, F., Mead, S.J.: Character-Based Interactive Storytelling. IEEE Intelligent Systems 17(4), 17 24 (2002) 9. Tychsen, A., Hitchens, M., Brolund, T.: The Game Master. In: ACM proceedings of the second Australian conference on Interactive Entertainment. ACM Press, New York (2005) 10. Delmas, G., Champagnat, R., Augeraud, M.: Plot Monitoring for Interactive Narrative Games. In: Proc. international conf. on Advances in computer entertainment technology, pp. 17 20. ACM Press, New York (2007) 11. Delmas, G., Champagnat, R., Augeraud, M.: Bringing Interactivity to Campbell s Hero s Journey. In: Cavazza, M., Donikian, S. (eds.) ICVS-VirtStory 2007. LNCS, vol. 4871, pp. 187 195. Springer, Heidelberg (2007) 12. Delmas, G., Champagnat, R., Augeraud, M.: Plot Control for Emergent Narrative: a Case Study on Tetris. Int. Jour. on Intelligent Games and Simulation 4(1), 29 34 (2008)