Narratives take place somewhere, in some world (real or imagined or blended). In many cases, the author provides just enough detail for his audience to follow the narrative. In traditional forms that worked well, sometimes even in a literal sense (e.g. sound stages filling in for vast empires).
But in modern media, this may not always work. In many cases, the narrative comes from the audience, and the author provides the tools and setting. Those open sandboxes allow entirely new narratives and play to emerge, often in directions the original author never intended.
So procedural generation can really open new doors, if you’ll pardon the pun. Creating a world based on an algorithm is an old trick, of course, from the days when programmers still worried about storage space and memory limitations. As a kid, I remember Starflight having fractally-generated planet surfaces, just to take one of many examples.
CityEngine released a demo reel a few years ago that starts to give some idea of what this can accomplish with today’s technology, where full cities create themselves down to the details of rooms and possibly NPCs.
More recently, they released another video showing the progress of the engine to date.
When I first watched this, I had a similar thought to Greg Smith’s over at Serial Consign:
What bothers me about the Structure demo video is how much it is tied to the first-person shooter genre – point of view is synonymous with wielding a firearm and the “destructibility” of matter is a major selling point... In the procedural cities generated by Structure everything just crumbles uniformly, and this is ultimately my confusion about Prototype – why go to the trouble of creating a complex, detailed, simulated city and neutralize it with gameplay that is entirely contingent on spectacular, unending mayhem?
This sort of technology creates tremendous opportunity for shared storytelling and exploration. Why must it focus on death and destruction? I’d really like to see this incorporated into virtual worlds, for example, particularly those not focusing on violent conflict. Maybe explorations of culture and lore, or economy and politics, or science and history could use procedurally-generated worlds to feasibly create large new spaces or represent possible ones.
Really, they’re thinking too small. What could you do with this sort of idea?