Prototyping The Prototype

This week, in my meta analysis class I continue to research Narrative Transport and find a large amount of evidence suggesting that Narrative Transport has a small-medium effect size on Narrative Persuasion. Then in my Research Through Design Class, I present two completed prototypes and a third prototype in development. Finally, in my independent studies, I begin looking into gathering more info on industry techniques, I describe my journey in developing a metric for quantifying believability, and discuss the next steps for Gaydar.

Meta Analysis

For this week, I continued to research the topic area of Narrative Transport's correlation with Narrative Persuasion and so far, apart from a single study there is a large amount of evidence to suggest that there is a small - medium effect of narrative transport on the persuasiveness of a narrative. Although it should be noted so far most of my research largely has been outside of the field of game design.

Additionally, I also was able to learn how coding sheets work and what it actually means to be a coder in a study. 

Research Through Design

For this week, I began exploring my first two prototypes for potential mechanics that could be used in Rae and I’s game. The first was mentioned last week of exploring creating a simple gossip system where NPCs would overhear what other NPCs are saying and then share that information with one another. Then for my second prototype, I began exploring creating a system to detect when a player performs an action and letting that modify what an NPC says. Below I go more in depth with the findings of each.

Prototype One: A Simple Gossip System

Description

The gossip system essentially acts as a way of tracking NPC knowledge on a local level (the knowledge is tied to a specific NPC rather than is something shared between all NPCs) and ways of passing that information amongst one another. This is meant to act like how people do not have a shared memory and instead rely on finding information out by either witnessing an event or by hearing about it from others. To handle the first scenario, the system also adds in an overhearing system so that nearby NPCs can listen in on a conversation. To handle the second scenario, the system allows for NPCs to have a routine which can include sharing information with others if they are nearby. 

Why Its Made

The purpose of this system was meant to help increase believability through increasing an NPCs awareness and social believability (Gnomes et al 2013). More specifically, through allowing the NPC to be able to know what is being said around them and keep track of that information and through helping the player “be able to identify social relationships between characters” through seeing them sharing information with one another.

How It Works

The way the system works is as follows:

1. The dialogue tree can contain markers that indicate what content is being communicated to the player

2. A dialogue system is to display the dialogue contained in the dialogue tree

3. An NPC tracking system that is able to determine who is near who

4. A schedule system that describes a routine with the following components; movement, idle, chit chat

5. The markers in (1) when reached are detected by the dialogue system (2), the dialogue system will broadcast that information to nearby NPCs from the speaker using the tracking system (3)

6. The NPCs when given new information will store it in their memory

7. When an NPC reaches a chit chat routine step (4), the NPC will compare their memory with those they are chit chatting with and update each other accordingly

8. When the dialogue system (2) reaches a branching point in the dialogue tree (1), it will check if the npc knows the dialogue flag (6) and play the corresponding node

How Did Players Respond

So far, players who have played the three versions of the game have universally stated that Version 3 with the full system is more believable than the other two versions. This gives strong evidence that the system is at least providing some benefit to an NPC’s believability.

Next Steps

The project still has the following gaps that will be addressed and iterated upon in future versions:

1. Tracking the relationship level between NPCs

2. Allowing the relationship level to change between NPCs

3. Tracking the secretness of a piece of gossip

4. Determining if a piece of gossip should be shared based on NPC relationships

5. Determining what piece of gossip to prioritize committing on to a player

6. A method of introducing false gossip (mishearing, misremember, misleading)

7. A method of introducing trust values based on NPC relationships

8. A method of stating who told the information

9. A method of allowing the player to overhear other npcs

Designer Commentary

Overall, I was quite proud of this first system. I decided to mainly focus on this basic test since I felt it would be the quickest to implement while still being obvious to the player. So far it showed pretty positive results but it's also revealed just how many more possibilities there are for this system.

One big surprise was that it introduced a level of tension to the game since you were constantly on edge about who is listening in on you and what might get leaked. This is actually quite exciting since it should add to the narrative game we are trying to create since in a way it adds a level of distrust to the social gathering as a whole.

Prototype Two: An Action Observation System

Description

The observation system represents a system that allows NPCs to be able to observe user behavior and then react in accordance to said behavior. In the first version, they can detect when a player enters an area and change their dialogue in response.

Why Its Made

The system is primarily meant to help improve the first prototype while further enhancing the awareness of the NPC (Gnomes et al 2013). Primarily through making the NPC be more able to know what is going on around them.

How It Works

The system is made up of the following components:

1. A group of observables that can broadcast when their corresponding action occurs

2. A tag that can be associated with an observable action

3. The system described in prototype 1

4. A field of view cone that detect which observables (1) are in range and can detect when an observable broadcasts an actions completion

5. A system for when the NPC detects an observable action (4), the tag (2) associated with the observable (1) can be logged to the NPCs memory described in (3) and used accordingly

How Did Players Respond

So far my tests have resulted in players viewing the prototype as more aware when it can see players actions then when it cannot.

Next Steps

Based off this early prototype a number of future directions can now be pursued such as:

1. Creating additional actions (ex. Picking up objects, Talking with NPCs, Interacting with things in the environment, etc.)

2. Creating a system of adding time information to the action (something may be okay before event X but not after)

3. Creating a system to modify an NPCs schedule based off newly obtained information

Designer Commentary

• How to prevent stealing spot light

• Connecting it to main story instead of feeling disconnected

• Spectrum of possibilities it allows

• Include testing details:

  • When, where, environment, task description, tester demographic information (?), Questionnaire

  • Spreadsheet recordings

Link To Survey: https://forms.gle/XEmE8yCVBskTsCoA6 

Prototype Three: Joint Prototype

After developing the two different prototypes, Rae and I decided to begin creating the first joint-prototype for the project which revolves around a simple series of scenes that goes as follows:

• Scene One: The player arrives at the party and chats with a few other party goers

• Scene Two: Everyone breaks off to go to their separate rooms of green and purple with the player being expected to go to their room, the green room. However, their companion brings up their desire to be with those in the purple room. Thus as a player you are left with the choice of which to join

• Scene Three: The room you enter reacts to your appearance and you are left with the consequences of your decision.

• Scene Four:  You exit the room as the party begins wrapping up and everyone is heading to the central area.

The main goals of this will be to explore the queer metaphor of selecting which room to enter and the gossip mechanics of people starting to spread rumors about you regarding which room you're entering and comments you're making about the rooms. 

Independent Studies

Gaydar

This week, Rae and I went about gaining approval from our committees that pursuing an IRB for gaydar is worth the effort. After a few discussions, we successfully managed to convince them and will now be moving forwards with designing the experiment / updating Gaydar in order to conduct the research.

GDC Talks

In order to better understand the field from an industry perspective, I decided to begin conducting a review of GDC videos related to narrative design from a systems perspective / NPC perspective. After reviewing all videos containing “Narrative Design”, “Narrative System”, “Telltale”, “Quantic Dream”, “Life Is Strange”, and “Dialogue System”, I managed to identify a collection of 40 videos that I will begin to watch and analyze over the coming weeks and then add them to my annotated biography of my research that I am writing. 

Believability

This week, I went about continuing to research various metrics used by researchers for believability with the main focus being on a criteria posed by Bogdanovych et al (2016) which itself was based off of a PHD thesis by Loyall (1997). Overall, I found the paper to be a great breakdown of different ways to improve NPC believability and liked the approach they took to weighting the different methods against one another to determine which makes the most impact. In the end, the two most crucial factors they found where to have NPCs have physical limits (cause it REALLY doesn't look realistic if an NPC can have infinite carrying capacity) and for the NPC to be aware of their surroundings and their role in the setting and how to accomplish their tasks. 

After going through and analyzing the three studies I’ve been looking at, I then went through and identified all the shared attributes between the criteria presented and created a unified metric that could be used for believability research. My goal is to now take that criteria and conduct a research study using the metric to categorize a collection of NPCs and determine if it's a useful metric for analyzing NPC believability to predict Narrative Transport. Additionally, to then narrow down the metric to a more refined list of questions that still maintains its usefulness. Hopefully, through doing so I will be able to propose a unified metric that can begin to be utilized in believability research in order for better comparison of NPC believability across papers while also providing a list of different points a developer could target when trying to make believable NPCs. 

Joint Thesis

For this week, Rae and I began working on our Joint Thesis Prototype as described above while also discussing our plans for our Joint Thesis and creating a contract between the two of us. 

The Wrap Up

Overall, I would say a large amount of progress has been made this week. Although I won’t lie, the prospect of now having potentially three different studies needing to be conducted is quite insane of a prospect to me, especially the part that each will need an IRB. However, I won’t have it any other way. 

Work Cited

Barreto, N., Craveirinha, R., & Roque, L. (2017). Designing a Creature Believability Scale for Videogames. In N. Munekata, I. Kunita, & J. Hoshino (Eds.), Entertainment Computing – ICEC 2017 (Vol. 10507, pp. 257–269). Springer International Publishing. https://doi.org/10.1007/978-3-319-66715-7_28

Bogdanovych, A., Trescak, T., & Simoff, S. (2016). What makes virtual agents believable? Connection Science, 28(1), 83–108. https://doi.org/10.1080/09540091.2015.1130021

Gomes, P., Paiva, A., Martinho, C., & Jhala, A. (2013). Metrics for Character Believability in Interactive Narrative. In H. Koenitz, T. I. Sezen, G. Ferri, M. Haahr, D. Sezen, & G. C̨atak (Eds.), Interactive Storytelling (Vol. 8230, pp. 223–228). Springer International Publishing. https://doi.org/10.1007/978-3-319-02756-2_27

Loyall, A. B. (1997). Believable Agents: Building Interactive Personalities.