Skye Winters
Project Conversation
Role: Lead Programmer and Level Designer
Team: Rae Watters and Skye Winters
Development Period: September 2025 - December 2025
Software: Unity and Arcweave
Core Questions:
How do we improve gossip systems to increase NPC believability?
How do we design systems that help NPCs be aware of their surroundings and reflect said awareness during dialogue?
How do we create systems of information sharing for NPCs in a way that can be easily implemented by future developers?
Description:
Currently, while several games have methods to have NPCs interact with the player, very few typically have ways for simulating NPCs relationships to one another. However, several papers have identified that social relationships can greatly assist in increasing NPC believability (Barreto et al 2017, Warpefelt and Verhagen et al 2017).
Additionally, it has also been found that increasing an NPCs awareness of their environment can further increase NPC believability (Gnomes et al 2013). For example, this is the difference between players being able to steal from a shopkeeper with a bucket over their head (see Skyrim) vs NPCs shouting out each other's names when they are killed (see the Last of Us 2).
That then raises the question, why increase believability? NPC Believability refers to an NPC "who seems lifelike, whose actions make sense, who allows you to suspend disbelief" (Mateas 1999). A key aspect of note here is the suspension of disbelief. The reason for this is that suspension of disbelief has been shown to help with the persuasiveness and enjoyability of a narrative (Van ‘T Riet et al 2018, Oh et al 2020). Furthermore, there is the possibility that believability could increase emotional attachment to characters. Therefore, for these reasons, to make more enjoyable and impactful games it's important to consider believability.
The role of this prototype is meant to serve as a case study of a potential gossip system implementation being developed by my thesis partner Rae and I. While several previous solutions have worked from the ground up defining a method for implementing a gossip system, our is instead a combination of a variety of papers proposals in order to take advantage of each their strengths (McCoy et al 2011, Kreminski 2023, Mooney and Allbeck 2014, Ryan and Mateas 2019). However, while creating this system, we realized that we would not have enough events to gossip about within the scope of the project and thus also decided to include a story generation system. This work is based upon the work of several past scholars (Johnson-Bey et al. 2025, McCoy et al 2011, Crawford 2004, Evans and Short 2014).
To evaluate their usefulness, we decided to create a short narrative that would last around 10-15 minutes. This also provided us insights gained from implementing the systems into a playable format. Throughout the process, we peformed user tests with the game to determine what aspects were effective and uneffective. Additionally, we maintained a development journal detailing the design challenges, insights, and decisions that were being made as well as knowledge gained through a complimentary literature review.
Overall, through this project, we were able to improve our gossip system and story generation system. Below we provide links to the development journal posts and 10 key insights from our research. Future work will include attempting to implement these systems into a new project that can better utilize their capabilities and then expanding the project to explore how well they hold up in longer gameplay experiences.
Methodology:
To see the process of creating these prototypes and developing this project, take a look at the following blog posts:
Update #1: https://skyewinters16.wixsite.com/skye-winters/post/the-start-of-another-journey
Update #2: https://skyewinters16.wixsite.com/skye-winters/post/the-hydra-that-is-a-thesis
Update #3: https://skyewinters16.wixsite.com/skye-winters/post/one-more-step-forwards
Update #4: https://skyewinters16.wixsite.com/skye-winters/post/prototyping-the-prototype
Update #5: https://skyewinters16.wixsite.com/skye-winters/post/good-old-research
Update #6: https://skyewinters16.wixsite.com/skye-winters/post/what-s-one-more-study
Update #7: https://skyewinters16.wixsite.com/skye-winters/post/back-to-the-drawing-board
Update #8: https://skyewinters16.wixsite.com/skye-winters/post/failing-quickly
Update #9: https://skyewinters16.wixsite.com/skye-winters/post/down-the-rabbit-hole
Update #10: https://skyewinters16.wixsite.com/skye-winters/post/reflection
Update #11: https://skyewinters16.wixsite.com/skye-winters/post/sanity-check
Wrap Up: https://skyewinters16.wixsite.com/skye-winters/post/wrapping-up
Takeaways:
The biggest takeways from this project were as follows:
When desigining a gossip system, you should make the exchanging of information visible to a player
When designing a gossip system, it helps to track speaker, target, event (past / present tense), related tags, anti tags, and an ID
When desigining a social media system, considerations should be made for not overwealming a player with seemingly unconnected information
When desigining a social media system, a mad lib style of writting can be useful when combining with a gossip system
A story generation system can be useful for generating gossip
When desigining a story generation system, considerations should be made regarding # of actions, making actions feel unique, not making actions feel repeated, and what the purpose of said system is
When desigining a story genreation system, a weakness of them is coherence between story beats and thus deserves more area of research
When desining systems, they should in someway benefit the players experience, either indirectly or directly such as through immersion or game enjoyment
To increase believability, story generation systems can provide NPCs with several aspects of believability
Story generation systems can benefit many aspects of games and should be considered in more applications. For example, background NPC actions, NPC party members during down time, narrative-driven games for replayability
