From January 20th to the 25th, my PhD supervisor Mike Cook and I ran an exploratory study on player responses to a “generative archaeology game.” The game in question is called Nothing Beside Remains-Mike originally made it in 48 hours for a jam. He’s very kindly let me work with it as part of my doctoral project, which has been fantastic for me as someone who is still new to programming. NBR is set in a ruined village with limited interaction mechanics; you can walk up to objects to see a short description. There is no real aim except to explore until you’re satisfied. Oh, and the village is procedurally generated. That’s pretty important.
We had participants play the game and then fill in a Google form, detailing their interpretations of what they thought had happened to the village. We also collected data on participants’ path through the village and what objects they interacted with. I can’t go into detail, but we were absolutely blown away by the responses. During the survey period several people expressed an interest in wanting to hear more about the context and results of the project. While an in-depth analysis likely won’t be published for several months, I can at least write this short blog post on the motivations and process behind the work.
Generative Archaeology Games
This is the first player study I’ve done as part of my PhD project, currently titled “Procedural Generation as Generative Archaeology.” Inspired by my background in archaeogaming (the archaeological study of games), I’m interested to see if I can get players to archaeologically interpret and record procedurally generated content. I would argue (and I believe the results of this survey back me up) that archaeological interpretation can be understood as a form of emergent storytelling, in which narratives about the past are created from limited material evidence. Archaeological recording can come in many forms-standardised written records, photography, cartography and more. By having participants interpret the village in Nothing Beside Remains, and write about it, I was hoping to explore the potential for interacting with procedurally generated content in this way.
This work has wider applications in procedural narrative design and games preservation-after all, the best time to preserve such an ephemeral experience is the moment in which it’s actually happening. If you’re interested to hear more about my research, I did a talk about object biographies and procedural generation at Roguelike Celebration.
For the purpose of our study, we chose two seeds to present to participants as a point of comparison. We deliberately chose two versions of the village that were strikingly different in their configuration to see how that might affect the responses we got. As you can see from the maps below, one seed had a much more regular grid-like pattern, and the other had a rather unusual layout with two distinct areas connected by a path.
“That Darned Sandstorm”
We soon realised that we should have play tested the seeds more thoroughly after participants reported a glitch that locked the game when they walked beyond the immediate boundary of the village. There is a glitch that we didn’t account for that locks the game when you walk outside the boundary of the village (this is actually in all iterations of NBR not just the seeds we chose). We could have just removed it from the build-but some participants were already incorporating what they called “the sandstorm” into their interpretation of the village and questioning if it was intentional or not. Not wanting to tamper with the experiment we had already set up, we ended up leaving it in, despite the glitch causing a lot of understandable frustration. If you were one of those lonely travellers who got caught in “that darned sandstorm,” I apologise. Your curiosity is appreciated.
There is so much more to say about that little village in the desert, but it’ll have to wait. Until then…
“The lone and level sands stretch far away”