A Fresh Start

In the following post, I discuss my four week progress review of Project: Theatre, the start of Project: R&J, and the updates to the thesis.

Introduction

This week has been eventful regarding my thesis. Specifically, as I mentioned last time, after experimenting with a simple prototype of the game using my system, I noticed several design flaws that would need addressing. After discussions with both my thesis partner and advisor, I’ve since decided that going in a new direction would likely be best. Specifically, I’ve decided to move away from the story simulation system towards a dialogue permutation system which varies dialogue based on the personality and relationships of a character. So without further ado, let's get into the details.

The Dialogue Permutation System

As you can see from the above, I’ve moved away from the name StoryGen to simply calling it the Dialogue Permutation System although I guess you could also call it DPS for short. Essentially the way it works is that it draws upon inspiration of traditional dialogue trees’ dialogue structure, salience based dialogue systems (Short 2016), the Praxis language from Versu, and Ortony (2003)’s writing on personality modeling. Specifically, it addresses thespian style problems where you don’t know the NPCs involved in a scene when it begins so you need to create a generic dialogue scene so that it can match any character. However, this can result in a loss of characterization as you can’t write to a specific character or a large amount of narrative burden as you need to write for every character permutation. Below I have diagramed an example of this type of situation.

To address this issue, I’ve decided to create a simple permutation expansion of dialogue systems using the Twee format where you can specify variations for a role’s dialogue line that is changed based on if it meets the specified requirements. These requirements are modeled by the Praxis language so that they can have a wide range of complexity. Using this, one can then create minor dialogue permutations based on aspects such as personality or relationship status. To illustrate this I’ve created the following diagram scene below.

The reason that I’ve decided to take this approach is for a couple of reasons

1. This allows for familiar style of writing since you can treat it as the equivalent as writing branching dialogue which presents the illusion of choice through characterization which matches the skillset of the people who would use this type of solution

2. This causes for a large amount of different potential capabilities with relatively little complexity since through using the Praxis language, entire world states can be modeled and tracked very quickly and easily. Additionally, you could make branches both on the in text level or as branching to alternate node paths depending on the depth of variation needed

3. Based on my research on believability, the core aspects are a) personality, b) illusion of life, c) emotions, and d) relationships. Through this system all but c) can easily be addressed through modeling memories, personality, and relationships using the Praxis approach outlined in Versu’s writings.

Thus for these reasons, this seems like a fruitful path to wander down to see what new discoveries can be made. The next steps for this will be to incorporate a way to update the praxis world state through the dialogue to finish the circuit.

Project: R&J

To experiment with this approach of DPS, I’ve began creating a simple visual novel based off the story of Romeo and Juliet where you play as the role of Mercutio or the pacifist who is intertwined with all the other cast members but must act from the sidelines as the tragedy of the story enfolds. Through using this premise, it gives a narrative reason for why the player has less direct influence and must instead focus on shifting the direction of the story through choosing who to play each role and pushing the story into situations that will unfold in a fruitful way. Additionally, this allows me to explore the aspects I want to address using this system in a pure dialogue format making it very applicable and easily demoed. Finally, this helps evaluate how well this works across a continuous story with the same cast of characters to test its viability for those types of stories.

Project: Theatre

Meanwhile, to test the viability of the system to work with the same cast of characters but in various different roles in short form events, I am working on the management-style game Project: Theatre. Since I’ve talked about it extensively in past posts, I’ll keep this one brief. The biggest updates were incorporating this new system into the events and polishing up the first demo level with the rehearsal system for use in playtesting. Additionally, after presenting this to several of my peers, I was able to obtain a large amount of feedback regarding the UX of the game. This was accomplished using the playtesting method of not providing any instructions / guidance to the participant so as to see as natural of an interaction as possible and identify what issues would occur if someone were to use the game when I was not present. I’ve already incorporated several of those tweaks with the next biggest one being the inclusion of a tutorial.

In addition, I’ve now begun maintaining a list of each of the features in the game and their corresponding design reasons in order to provide clear documentation in my thesis for why each element was incorporated. Below you can see the current version of this list. 

Label	Description	Reason
Time	A time pressure system is incorporated within the rehearsal system. The timer lasts 3 minutes and is paused when in narrative scenes or assigning to events but not when looking at the environment map nor on a character assignment page.	Helps prevent analysis paralysis which is when players try to “act perfectly” due to not having the time to “act perfectly” Creates tension as the player balances optimization with threshold for what they can manage which matches the story Helps introduce a flow state to the player since it requires the player to focus or otherwise they will fall behind
Casting NPCs	Instead of a fixed list of characters that will occur in the story and what roles they play in the story, the player instead is given the choice of casting the characters in whatever role they wish.	Increases theming since your role is a stage manager in a stage production Allows for a built in story reason for why my thesis is needed since we don’t know what characters are in the scene in advance Allows for greater replayability and customization in the game
Events (Static)	At fixed points within a rehearsal, an event will appear requiring the attention of the player and the player assigning 1-3 cast members to solve the issue. Each event has a set of skill requirements and a corresponding dialogue scene upon completion.	Adds variation to the gameplay loop Presents an in game reason for why my thesis is being used since these are the dialogue moment mine is being incorporated into Presents moments to highlight the personality of different cast members
Events (Dynamic)	When specific requirements are met during a rehearsal, an event will appear requiring the attention of the player and the player assigning 1-3 cast members to solve the issue. Each event has a set of skill requirements and a corresponding dialogue scene upon completion.	Adds variation to the gameplay loop Presents an in game reason for why my thesis is being used since these are the dialogue moment mine is being incorporated into Presents moments to highlight the personality of different cast members
Task Assignment	The player during rehearsal is given the task of looking at what needs to be completed both on an individual level and a production level and delegating cast members to complete those objectives.	Provides the central gameplay loop by leveraging past games for examples (specifically the management game genre) Thematic since the player’s role is a stage manager so increases the player’s immersion through alignment with the protagonist they are playing as.
Full Game vs Single Level	The game is being designed to be a full 1-2 hour long narrative experience rather than a 20 minute prototype.	Allows for evaluation of the game system within more realistic game constraints. Specifically, how does it hold up when narratives include multiple encounters over the course of several hours rather than a few minutes. Serves as a good portfolio piece upon graduation
Scoring System	The game scores the user based on the management of events, individual tasks, and production tasks.	Presents an incentive for the player to focus on both the assignment phase and events Presents an interesting cost problem for the player of balancing tasks and events since an NPC can only be doing one at a time
Calendar	The game occurs over the series of several days within a single month. Specifically, 11 different rehearsals plus the intro scene a few weeks prior.	Helps the story feel like its progressing somewhere Helps explain why the characters are able to form relationships with one another over the course of the story Helps maintain the theming of being a stage manager putting on a production by following a production timeline
Character Customization	The game features the ability for the player to customize the character they are playing as.	Allows for the player to feel more invested in their character since they are helping contribute to who they are playing as. Provides support to Rae’s thesis since character customization is part of his relationship calculation system
Tutorial	The game opens with a tutorial explaining the game mechanics to the player	Allows for easier playtesting since game can just be given to testers rather than the developer guiding them through the experience Necessity for the game since its going to be made publicly available for people to download and play
Twine Tool	The game makes use of the twine dialogue tool for creating both Rae and I’s thesis dialogue systems.	Twine is open source so therefore we can open source our systems for others to expand and use Allows us to not have to recreate a dialogue system from the ground up which would take away time from the game’s development Allows us to expand from past twine based games we have made
Storylet Inclusion	The game features storylets based design as shown through the dynamic event system. Storylets are snippets of stories that have prerequisites for appearing and post conditions for modifying the world state upon completion.	Storylets allow for creating stories where you don’t have a fixed order in which the story moments will unfold. Allows for a unique playing experience since features greater reactivity to player action through prerequisites being tailored to past player actions
Personality Inclusion	The game features a dynamic narrative variation system during events where it will present minor variations based on the NPCs personality. The NPCs’ personality is based off the Big 5 Model of personalities	This is the core of my thesis so kinda important to include Uses the Big 5 Model since its a very well respected and well known method of personality modeling that captures a wide range of characteristics Helps allow for maintaining characterization in events where the speaker is unknown to the story author
Character Relationships	The game features a dynamic narrative variation system during events where it will present minor variations based on the NPC’s relationships. The NPCs’ relationships are tracked through state variables and past memories	This is also core to my thesis Uses memory to give the illusion of life and relationship status since it provides modeling social relationships, both of which are core aspects of believability Helps prevent incongruency in the story where the NPCs act in ways that don’t make sense given past situations they encountered when the story author doesn’t know what the characters have been through till run time
Day Time System	The game features a day time system in the style of traditional visual novels where players’ are able to bond with other members of the cast.	This allows for separation between Rae and I’s thesis since he has the day and I have the night This allows for a very narrative driven experience which allows for leveraging Rae’s skill of writing Presents a contrast to the management style gameplay in the night to add variation in a way similar to the highly successful game Dispatch
9 Castable characters but 6 used	The game incorporates 9 different NPCs for which you can cast but only 6 of which you can cast.	Presents replay value from having content that is guaranteed to not be incorporated in the first game. Allows for a wide range of personalities to be modeled to see how well both characters large amounts of similarity and little similarity function in the game

Overall, the next steps now is further refinement starting with the creation of a tutorial now that I’ve identified which aspects of the game are the most unintuitive to users. From there, I’ll work on expanding the content to include more events in the tutorial scene to further explore this solution across events. Finally, I’ll create a version of the game where it auto selects a set path for each character and then I will compare it to my system to see if my solution is providing an increase in characterization of the NPCs. This will be the first step in determining the usefulness of this approach. Below, I’ve included the four levels of success of the system which the goal to hopefully achieve a level equal to L3.

Believability Study

Finally, regarding the believability study, I’ve now gone through and created the third version of the believability questionnaire and have continued to look for more stimuli to give to participants. Apart from that, not much progress this week since the lead advisor was away this week at conferences.

Conclusion

Overall, I would say this was a very productive week in terms of progress. I’m looking forward to seeing where my thesis heads next. 

Until next week, logging off.

Citations

Evans, R., & Short, E. (2014). Versu—A Simulationist Storytelling System. IEEE Transactions on Computational Intelligence and AI in Games, 6(2), 113–130. https://doi.org/10.1109/TCIAIG.2013.2287297

Ortony, A. (2003). On Making Believable Emotional Agents Believable. In R. Trappl, P. Petta, & S. Payr (Eds.), Emotions in Humans and Artifacts (pp. 189–212). The MIT Press. https://doi.org/10.7551/mitpress/2705.003.0007

Short, E. (2016, April 12). Beyond Branching: Quality-Based, Salience-Based, and Waypoint Narrative Structures. Emily Short’s Interactive Storytelling. https://emshort.blog/2016/04/12/beyond-branching-quality-based-and-salience-based-narrative-structures/