Abstract

In this thesis we take a closer look at Tutorials in video games and Tutoring in general. A Tutorial can be seen as a closed environment, where we have total control over the challenge and the skill level of the player. We are therefore looking at the role of Tutorials in the Flow theory by Mihaly Csikszentmihalyi. We define an optimal amount of knowledge, that should be taught by Tutorials, which lies between an explicit minimum and an implicit maximum. The minimum can be taught with a Mandatory Tutorial in the beginning of a game. Additionally we have implemented an adaptive system, which provides Hints to the player depending on the individual performance, such that experienced players still receive valid Tutoring. There are many ways to track the player based on ingame variables and the right visualization of the data has proven to be a reliable tool for self-driven improvements. However the resulting Hints, which can be seen as an additional source of Help, did not adapt to the player as much as it would be desirable, due to several reasons. We have implemented all described Tutorial solutions in the highly challenging game "Afterthought" programmed by me and my Team at Studio Moondowner.

Research

Research Question 1: What is the role of Tutorials in the Flow theory?

We have stated, that Tutorials needed to increase the player's skill level while confronting the player with appropriate challenges. Therefore we have modeled Tutorials as closed environments, where we have total control over the challenge and the skill level of the player. On the example of the Mandatory Tutorial in Afterthought we have explained in depth how we can teach game mechanics to players, while engaging them with increasing challenges. Especially during the final challenge of the Tutorial the player is likely to enter a Flow state, as the skill level, which is needed to clear the challenge, is equal to the minimum skill level needed for a Flow experience.

Research Question 2: What is the optimal amount of knowledge, that should be taught inside Tutorials?

We have proposed, that an optimal amount of information lies between an explicit minimum and an implicit maximum. The research model in the figure below has mapped the minimum and the maximum onto the skill axis of the player. According to our theories the minimum is closely related to the first, mandatory Tutorial, and corresponds to the minimal level of skill which is necessary in order to enter a Flow experience. The maximum depends mostly on the developers, who need to make a decision. Everything which is not taught in Tutorials is considered to be a hidden mechanic. However the existence and the interactions of hidden mechanics can be shown implicitly through unambiguous ingame feedback. Beyond the maximum set of information a skill ceiling is supposed to exist, which represents the highest possible skill level, that can be reached inside a game.

Research Question 3: How can we tutor an experienced player?

We expected that experienced players would need less guidance by explicit Tutorials and more guidance from implicit Hints. Anyway the Hints have turned out to be less powerful than the simple visualization of the last playthrough. We have built the Help Screen as a Tutoring tool for experienced players, which confronts players with their mistakes and improvable ratios. From testing we have seen, that inexperienced players did not benefit from the information as much as players with more practical knowledge did. The visualization of the acquired data during the last playthrough has proven to be a reliable tool for self-driven improvements. We see, that experienced players are more thankful for the raw data than for the Hints, when they are seeking for information on how to perform better. However due to organizational problems the feature could not be properly tested on professional players, for whom the system could have unleashed its full potential.

Research Question 4: How can we utilize an adaptive system in order to tutor the player?

We have stated, that an adaptive system could measure the player's performance in order to determine the right type of Tutoring. The Hint Decider on the Help Screen has been built as an adaptive system in the way, that it selectively chooses from a set of Hints based on the player's performance. Testing has shown, that the measurements could model the player very accurately. On the other hand in a game with so many possible interactions, we could not distinguish between the states indicated by the data very well. In some cases (especially in short levels) small outliers in the player's performance have led the Tutor in a wrong direction, such that confusion and misinformation have been spread by the Hints. In other cases the Hints were simply too general and did not adapt to the player that much. Due to several reasons the adaptive system discussed in this thesis has partially failed, however a second iteration based on the research might be able to adapt to the player in a far more reliant way. Further investigation is needed in order to optimize the system for the commercial product.

Feel free to play the Demo on Steam in order to test the Tutorials.

Files