By Bjarne Kristiansen & Oluf Kjærgaard Pedersen
Game Research Cluster, ITU Spring 2007
Supervisor: Jesper Juul
The art of conversation 1
Original Method 4
Revision of the problem and a new method 5
Presenting the previous work 6
Continued analysis of games and conversations 7
Gothic 3 (2006) 7
Consequence versus inconsequence in conversations 9
Consequential conversations 10
Non-consequential conversations. 12
So what does this talk of consequence mean? 14
Connecting the dots 18
Critique of method 21
The Questionnaire 22
Results- Quantitative 25
Law of the West 27
The secret of Monkey Island 27
Planescape: Torment 27
Baldur’s Gate II 28
Vampire: Bloodlines 28
Neverwinter Nights 2 29
Gothic 3 29
Playing style summation 29
Quantitative comments 32
The goal of this project was to find ”best-practice methods” for conversations/dialogue options in games. This was to be accomplished through an online survey (quantitative) and follow-up interviews (qualitative), where playing styles and questions pertaining to conversations in a list of games were to be compared.
The quantitative survey (116 participants) did not yield indisputable proof of a general correlation between playing style and preferences in conversations in games, but did reveal a game-by-game difference, yet the three qualitative interviews suggested that a certain playing style did indeed affect a player’s preference in respect to conversations/dialogue options in games.
In respect to best-practice methods the overall findings were that future conversation/dialogue options designers take a long hard look at older titles such as Fallout and Planescape: Torment for inspiration, as these games were clearly the best at giving the player a good experience in regards to conversations and dialogue options.
The overall goal of this research project was to find “best-practice” methods in respect to conversations in computer games. This was to be accomplished by a thorough investigation into what effect dynamic/parametric conversation had on the experiences that players reported and if this was related to their player type, as originally described by Richard Bartle (Bartle; 1996).
We were building on a previous study into dynamic conversations made by one of the members of the group (see Appendix A).
We were also very interested in whether or not dynamic/parametric conversations in games were necessary or indeed even needed in games, because if not, then obviously there would be no real best-practice method except leaving them out or making them extremely simple.
The methods we would employ to try and answer our questions were one or both of the following:
We would choose sequences from games that use conversations, and observe player behaviour.
We would create our own "house of conversations" with varying degrees of conversation leading to the same game goal. Again with us sitting on the sideline observing.
These methods would then be followed up by interviews.
After several discussions in the group, it became obvious that neither of our methods would yield the results we were looking for, or were indeed productive considering the time and energy (and money) we would have to spend.
Having people playing through a particular segment of a game and observing them, would primarily yield results concerning that particular game, and if we wanted broad results, then we would have to have a lot of people playing a lot of games, with us observing. We had neither the time nor the money to conduct such an experiment.
So basically this method was abandoned due to us wanting broad results.
The second method of using a level-editor to create a “house of conversation” was also abandoned. We had planned to use the Neverwinter Nights (Bioware; 2002) editor, but decided against it, because the results would be very predictable, because this editor, or indeed any level-editor, would always be limited game-play wise to the original game-design. And since the original game-design in Neverwinter Nights placed an emphasis on conversations, it would be nonsensical to use it to make a level without conversations (or with very few). The results would be obvious: up to a certain level, the more conversation a level had: the better.
While it might have been fun exploring what the optimal level of conversation was in Neverwinter Nights (also in respect to playing style), it was again not as broad an investigation as we would have liked. And also this research had very likely already been done when the game was originally play-tested.
Revision of the problem and a new method
We slightly revised our problem: originally we were looking at the player-types according to Richard Bartle, but this categorization was based on behaviour observed in a multiplayer context, and we were dealing with single-player games. Our supervisor suggested that we use the player-types put forward in the book 21st Century Game Design (Bateman & Boon; 2005), and we took that to heart. These player-types are based on a psychological test called the Myers-Briggs test (Myers, McCaulley, Quenk and Hammer; 1998).
The new method we decided on was to conduct an online survey exploring peoples experience with conversations in games, and also to see if there was a correlation between their respective player-type(s) and their preferences concerning conversations in computer games. The survey would ask questions regarding the games analysed in Appendix A.
This survey would then be followed up by a smaller survey wherein we would interview people about the possible results gained by the online survey.
We would also be expanding the analysis regarding games and conversations begun in Appendix A.
In this report we will often be using certain terms. Here is an explanation detailing our perception of these terms:
Game-play: to us, this word means simply what the game mechanic is about. It does not concern the “story” of the game. An example: the game-play in PacMan (Namco; 1979) essentially consists of trying to eat all the cheeses while avoiding the ghosts (reflexes, coordination and thinking on your feet).
Dynamic conversations: a dynamic conversation is one that changes according to player choices in the game-world. So if for example I retrieve an object, X, and this object in my possession triggers person A to respond differently, then this is a dynamic conversation. It also follows that a conversation that never changes is linear. Law of the West (Accolade; 1985) is a game that employs a completely linear dialogue tree.
Progression: we take this word to mean simply moving forward/advancing in a computer game. This obviously doesn’t explain how this forward movement occurs, whether it be completing quests and gaining experience points or reaching a personal goal like being able to wield two swords simultaneously.
Immersion: we adapt the understanding of the word as put forward in Fundamental Components of the Gameplay Experience: Analysing Immersion (Ermi & Mäyrä, p. 4, 2005).