Entertain this juxtaposition for a moment. On the one hand, you’ve got ideas. Big ideas —themes rolling around your head so heavy, so important that you just can’t wait to share them with your players. On the other hand, you want to embrace the full range of interactivity that the medium of games affords you, and (for you) that means using generative systems to produce unbounded, novel, exciting, and surprising content.
I’m of the mind that you can do both of these things (and that they’re both worthwhile things to do) by embedding your themes in the hand-crafted units that your algorithms affix together. The collaborative roleplaying game Fiasco—which relies on procedural generation via pen & paper—does this tremendously well; its theme of ‘powerful ambition and poor impulse control’ manifests itself in the carefully hand-crafted units that players then algorithmically combine in the Set-up and Tilt (when you choose a Relationship, you don’t choose from among the full range of human relationships; no, you choose from a few such relationships, like ‘con-man’ and ‘mark’, that are inlaid intrinsically with the theme of ‘ambition’).
By outlining the storyspace in this way, we’re able to let our players tell their stories while exploring our themes. Wonderful! But Fiasco is relatively simple. How do we build complex generative systems with meaningful cores? How do we reap all the benefits of procedural generation while constraining the system to sufficiently invoke a vision, make an argument? Even more fundamentally, how do we talk about this stuff? Can we construct a vocabulary to articulate the dimensions of content generation?