Post by Rose Lazuli (Naitoshadou)

The Problem

When the team returned to My Ordinary AR Life this semester, it became quickly apparent that the scope was far too large. This included the narrative side of things, and as such, it was decided to scale down the narrative to just one character, the Aunt. Doing so would allow for player connection to exist, as spreading it out over too large a cast would leave players feeling overwhelmed.

 But without multiple characters to play off of, it becomes harder to tell a story with multiple viewpoints that can inform the player. The Aunt could effectively show why AR addiction might be bad for the player character, but it was hard to show any other viewpoint. So a new model was needed.

The New Model

The new model is instead a series of short texts received by the aunt character. The player is never given a chance to respond directly, but might choose to modify their behavior, such refusing to use the in-game AR headset. Another major factor of the new design is contextual meaning: carefully writing the aunt's lines so what the player takes away is dependent on their actions, without changing any dialogue based on in-game actions. This is intended to drive the player in a certain direction by having them experience a sense of guilt or pride based on the choices made prior.

The Seed

This new model might suit the game's scope better, but it still needs a new scaffold and structure to inform the narrative arc. With a short timeframe, the arc needed to be direct and distinct, so that progression was felt clearly. This eventually led into some research into addiction, behavioral psychology, and the reasons people change behaviors, eventually finding the transtheoretical model of change This provided a rough outline  I could use to direct the growth of the narrative, acting as a core theme from the development end.

The transtheoretical model consists of five stages, which suits the length of the game well, focusing the narrative arc to five segments. Each stage is also well defined by certain patterns and emotions, which provide strong inspiration for writing characters.

The Dialogue Draft

Writing dialogue snippets ended up being a rough process. The new model, while functional, meant adjusting to a new and unfamiliar workflow. Combined with some stress-related motivational issues, this ended up making it hard to write dialogue directly.

As part of my process, I decided to focus on how the Seed could impact the Aunt's perspective directly, resulting in my roughing out some very basic emotions and motivations for the character at each stage. From there, I also tied each stage to a more common narrative progress structure: conflict-based tension. By determining how much conflict there needed to be to get a compelling feeling of conflict between the player and aunt character, it helped to outline how antagonistic she needed to be in a given moment.

Testing the Draft

Due to not being familiar with this model, and feeling the limitations of writers block, it felt necessary to test the early work to learn if I was using the model effectively, or if I needed to try something else than my current work. To help inform this, I derived 3 areas I felt were important to the dialogue's role in the game: the feelings it communicated and evoked for the player, the themes it communicated, and the quality of the writing. For each of these sections, I devised a test I could use to learn what was working and what wasn't. I put these tests together on a Miro board for testing using a sample audience.

Feelings

For the feelings, I set up a chart with two axes: what the tester felt the Aunt was feeling, and what the tester felt after reading the line. This would allow me to get a rough sense of whether the overall conflict was being represented well, with the player in the shoes of their character to feel the conflict and tension ramp up.

Themes

For the section on themes, in the interest of not driving the tester to any conclusions, I set up a simple collection of sticky notes the tester could fill in with a theme they had identified. They would then group the dialogue into the themes they had identified, informing me of both what they perceived and why in a simple and intuitive way.

Quality

The third test was similar to the first, but only used one axis. Testers simply ranked each dialogue snippet's quality on a scale of bad to good. This is intended to inform me which dialogue  isn't compelling or doesn't have good voice, allowing me to prioritize rewrites of certain lines and focus on what is working instead,

Other

The final section was just a field where the tester could leave any additional thoughts. These sections combined allowed me to focus on areas I felt concern about in the narrative and receive simple yet instructive feedback.

The Results

Feelings

For the first test, an interesting trend emerged: the more positive the aunt character seemed, the more positive the tester felt after reading the line, and vice versa. This means that the contextual shift in feeling and meaning might be difficult to communicate effectively. However, the overall trend of the positive and negative emotions seemed to be in line with the intended tension, which means that the flow is working in a way that can be followed up and improved upon.

Themes

In terms of themes, testers consistently identified a theme of loss of connection, in particular with family. This is one of the major themes of the aunt character, meaning that at least the character's intent is coming through well. However, addiction went unmentioned, meaning it's not currently apparent in the current dialogue. I'll need to review and see if this theme is communicated better with game context or if it needs to be more explicit to the dialogue. There's also potential here for a tester bias, as testers were young and technologically-minded, which may have them viewing AR in a more positive light than anticipated.

Quality

Testers overall grouped most lines into the mid-high sections of quality, indicating that, generally, the writing is effective. However, longer lines with more generational slang tended to be ranked lower. A shift in the aunt character away from inflections like "dearie" and "doohickey" may be helpful in communicating with a younger audience better, as they seemed to find the lines clunky.

Other

Testers overall liked the dialogue when it showed more direct interplay with the characters, but found elements designed to communicate the aunt's age either confusing or annoying. Some testers could identify the shift in attitude over time, while others felt like the contrast made the characters feel inconsistent. Testers in particular liked lines with common sayings like "right as rain" or lines talking about common elements of generation gaps and technology, meaning those areas were relatable and effective.

Conclusion

Overall, the structure is working, and the seed is informing the tension properly. However, the secondary intent of having contextual meaning seems like it may not work out long-term in development and should be reconsidered. Additionally, testers preferred snappy, shorter lines with dialogue elements and themes they could relate to. Major areas of concern are in identifying the flow of conflict, and the lack of addiction theme identification. The former is hard to draw conclusions on due to the testing method used, and might be worth focusing on during a test of dialogue in-engine. The addiction element indicates adding more explicit mentions to addiction in some way may be effective. Should test if additional context from gameplay fills in gaps or if it needs to be made explicit in coming weeks.