Divine Affaire

Divine Affaire is an experimental emergent-narrative game developed by Jonathan Lessard and LabLabLab. You take on the role of a god of a procedurally-generated world, attempting to answer the prayers of your believers.

Because of your divine omnipotence, you are able to see a complete, wiki-style record of the lives of each and every person in the world. This omnipotence isn’t perfect, however. The game’s systems can complicate seemingly simple interactions in interesting ways, and even a god’s plans can go astray. For every action, there is an unintended consequence, and your believers may well wind up wondering why their prayers for aid have gone so cosmically wrong.

Because of its innovative narrative-generation systems, Divine Affaire was also used as part of original research into narrative eventfulness, which I conducted alongside Prof. Lessard. Divine Affaire has been exhibited as a narrative experiment at ICIDS 2023 in Kobe, Japan and FDG 2024 in Worcester, Massachusetts.

My involvement in Divine Affaire is broadly divisible into two principal responsibilities: that of a games researcher and a systems designer.

Games Resarcher
As part of the LabLabLab team, I assisted Prof. Lessard in developing experiments in narratology to be conducted using Divine Affaire‘s systems. This entailed reading literary analysis what made stories in general noteworthy as well as contemporary games studies research on emergent narrative. We developed a system to filter events in Divine Affaire using Wolf Schmid’s criteria of Eventfulness. I then conducted experiments in Divine Affaire using these criteria, found stories which scored high or low on our chosen metrics, and converted these stories into AAR-style short stories. Participants were then asked to read these stories and evaluate them for interest, coherence, and plausibility. To control for authorship effects, each story was generated twice – once by me personally and once using ChatGPT 4.0. Though it was not the focus of our research, to my immense relief the human-authored stories consistently outperformed the GPT-generated ones.

More significantly, we found that creating systems to look for and surface events which fit established narrative patterns and which satisfied one or more of Schmid’s criteria resulted in more engaging stories on average than randomly-selected stories, suggesting an interesting avenue for the development of emergent-narrative story sifting systems.

Systems Designer
My work as a systems designer focused on building new kinds of narrative into the game’s systems, both personal and sociological in scale. I produced design documents outlining these intentions:

• Unhappy Families, which would expand the consequences of a character reaching 0 or negative Mental Health
• Trade update adding new jobs and needs
• Clashing Clans, detailing inter- and intra-community conflict
• The Nature of States, a system for creating stories of early political community and cultural schismogenesis.