Back to the Drawing Board

Back to the Drawing Board

This week, Rae and I encountered a few setbacks and decided to start fresh with the knowledge we gathered in our project. I also completed a few more readings for my thesis.

Research Through Design

For this week, Rae and I performed the first round of testing for project conversation. In the end, we had four different users go through the experience and received mixed feedback based on post gameplay discussions and our google survey described in last weeks post. The largest issues we concluded after our testing were that

1. Players were struggling to move around the environment

2. Players were getting confused about the story and / or felt the trans analogy was heavy handed

3. Players were not noticing the gossip feature

Based on this feedback, we realized that we would need to do some revisions since both of our thesis topics were the weak points of the experience. Thus, we decided to implement the following changes:

1. We would do a complete revision of the story line so that the analogy would be stronger overall

   1. Reason: Due to the diagnosis that the problem with the story was fundamental to the concept and not something that some light revisions could fix. Additionally, neither of us were particularly passionate about the direction it was heading.

2. We would remove all locomotion from the experience, making it largely visual novel style gameplay based primarily on Stray Gods

   1. Reason: The goal of the project was not to implement a 3D locomotion system so to minimize unnecessary development, we would remove it to a far simpler and clear solution. Additionally, it would better convey that you could / should travel to different areas between conversations.

3. We would move the game from a 3D style to a 2D style

   1. Reason: Due to both of us being primarily 2D developers from our prior game projects, we decided to play to this strength by using a 2D style game that is mainly UI based.

Since then the following updates have been made to the game:

1. A title screen and end screen have been created for improving game flow

2. The unity project has been cleaned and outdated scripts have been removed

3. The gameplay has been fully transferred to the new system

4. The new story has gone through several revisions before writing any dialogue to allow for a stronger foundation

   
   

Screenshots from New Version of Project

Moving forwards we will be looking into exploring the questions of:

1. How do you communicate to the player that the NPCs are gossiping about events in the game?

2. What actions should be captured by gossip?

3. How do you effectively represent the bonds between NPCs?

4. How do you make a gossip system into an entertaining gameplay mechanic?

Literature Review

In addition to the work being done towards our thesis project, I have also continued to review the research being done in the world of Social Believability. Through doing so I discovered the following papers:

Gameplay Design Patterns for Believable Non-Player Characters

In their work, Lankouski and Bjork (2007) write about the core components they feel are important for creating believable non player characters. For their understanding, they focus primarily on looking at believability through the lens of cinema studies as well as Daniel Dennett’s “Stances”. In the end they concluded that the game design patterns most beneficial are having initiative, an awareness of their surroundings, their own agenda, a sense of self, emotional attachments, contextualized conversational response, continuous goals, a goal driven personal development, and an open destiny.

The key impacts to my research are primarily the key components that they mention being important to NPC believability. Additionally, the technique of using gameplay design patterns to describe the components of a gameplay system.

Rethinking NPC Intelligence - A New Reputation System

In Mooney and Allbeck’s (2014) paper, they describe an alternative way of approaching a reputation system and how one could define it. Basing their work on the Belief-Desire-Intention (BDI) model, they describe how developers could use an RDF triple structure with the components of Subject, Verb, and Direct Object to represent a piece of information. Then for any given piece of known information by an NPC, they can have a float value representing how much they trust the information. Finally, for each piece of information, you could attach relevant tags such as “physical” or “harmful” to allow for further understanding of what an NPC knows. Then to share the information, you could have NPCs gossip with one another or hear others gossiping with one another. Then using that information, the NPC could use a Bayesian Network model to make future decisions.

Overall, the main take away from their work I will be incorporating into my own work is their method of representing information since I feel it is quite powerful while remaining relatively simple. As such, for our new version of the project, I’ve gone through and created an excel document that allows us to track every piece of gossip that can be generated during dialogue.

Sunset Valley: A Case Study in Computational Design

In their paper, Liao et al (2013) present a gossip system from the lens of sociology. In their approach, they focused primarily on the aspects of personality (based on the big 5) and relationship strength (positive vs negative). Then based on these two, they suggest applying a weight to information to determine the likelihood that the info is shared.

While not yet planning to use their work, they do bring up an additional approach that may be utilized in the future of giving the NPCs a personality makeup and then using said personality to alter gossip behavior. Additionally, they provide a good foundation of how sociology views gossip and why it's important for peoples’ well-being.

Toward a better gossip system

In their paper, Kreminski (2023) provide a breakdown of the current weak point in gossip systems is that majority of the work being done only focuses on some but not all of the following aspects:

1. Gossip is related to actions taken rather the basic information

2. Create and convey microstories about these actions in ways appropriate for the narrator of the gossip

3. Allow different characters to interpret and react in their own individual ways based on their own current knowledge about the subject of the gossip.

Additionally, they provide the following key phases for gossip to occur for a given NPC:

1. Witness Phase

2. Reflection Phase

3. Propagation Phase

4. Decay Phase

Finally, they also provide a description of a heuristic that can be used for Salience of information. More specifically, if a character knows a lot of information about another character, they would likely care more about that character and thus would be more receptive to future information.

Overall, several key takeaways can be reflected from this paper. The most obvious being a new criteria that could be used to access the quality of our gossip system. Currently, we are effectively doing (1) and beginning to work on (3) but have yet to approach (2). As such, we have now discussed potentially including a social media system similar to Prom Week. Through doing so, we would be able to have NPCs convey what information they know to the player while also not breaking players suspension of disbelief through doing other more gamified methods. Furthermore, we will also potentially use their method of salience if our current solution of relationship strength does not work out how we hope.

The Wrap Up

Overall, this was a rougher week with a number of setbacks resulting in not much time for work outside the work on our thesis project. In the coming week a lot more should be able to be achieved.

Work Cited

Kreminski, M. (2023). Toward Better Gossip Simulation in Emergent Narrative Systems. 2023 IEEE Conference on Games (CoG), 1–4. https://doi.org/10.1109/CoG57401.2023.10333140

Lankoski, P., & Björk, S. (2007, January 1). Gameplay Design Patterns for Believable Non-Player Characters. DiGRA Digital Library. Proceedings of DiGRA 2007 Conference: Situated Play. https://doi.org/10.26503/dl.v2007i1.262

Liao, F., Gelardi, G., Mitchell, K., Sandhu, A., & McCoy, J. (2023). Sunset Valley: A Case Study in Computational Gossip. 2023 IEEE Conference on Games (CoG), 1–5. https://doi.org/10.1109/CoG57401.2023.10333161

Mooney, J., & Allbeck, J. M. (2014). Rethinking NPC Intelligence—A New Reputation System. MIG ’14: Proceedings of the Seventh International Conference on Motion in Games, 55–60. https://doi.org/10.1145/2668084.2668091

https://store.steampowered.com/app/1920780/Stray_Gods_The_Roleplaying_Musical/

https://promweek.soe.ucsc.edu/