18 Cadence is a new piece of electronic literature that’s almost definitely not a game, something less than a book, and explores a rarely tackled corner of interactivity in interactive narrative: the choices and decisions of how to tell a story, what bits to include, what to leave out, how to arrange them. It’s available for free on the web or as an iPad app.
I’ve already written elsewhere about the piece generally, but I’d like to speak here to why I created a project like this as a computer science PhD student in a lab focused on something called “Expressive Intelligence.” The easiest answer is that I started the project when I was still a freewheeling hippie digital arts student, and am only just getting around to finishing it now (perhaps not coincidentally, I’ve just finished my core classes, too). But as a graduate student I’m used to delving deeper, so let’s delve.
During my MFA I was interested in moving away from the relative safety of parser-based interactive fiction, where I’d done some reasonably successful projects, and exploring new ways of making narrative text interactive. Now that I’m in a computer science program, I’m tightening that focus to be more explicitly the ways that generative methods and interesting processes can inform an authored narrative, but looking at things from the perspective of an artist (wanting to create an aesthetically pleasing artifact) rather than just a system-builder (wanting to create an aesthetically pleasing machine). If I make interesting machines along the way, I’ll be pleased (and so, I imagine, will my eventual dissertation committee) but it’s not really my primary goal. This puts me in a different space than many people working in interactive narrative, and also helps keep me sane as an artsy person in an algorithmsy environment.
Last year I made a short story called Almost Goodbye that minimally explored being both an interesting story and an interesting machine, but at first glance 18 Cadence seems to be neither of these things. It gives you a hundred years of a house’s history: thousands of fragments of events, objects, locations, and ages, but the job of assembling them into a narrative is left to you. (The granularity is smaller than in an avant-garde novel where you might assemble the pages in any order you like, but there’s less structure than a hypertext fiction with a fixed number of links and possible connections.) Not only is it a DIY story, there isn’t any fancy narrative AI system under the hood: no simulated environment a user can become embodied within, no algorithms that attempt to understand story structure or reason over character motivation. There’s a certain amount of lexical logic involved in making the fragments fluidly combinable while retaining their grammatical correctness, able to adjust their pronouns based on proximity to each other and so forth, but that’s more or less it.
Are there expressive processes, though? In a sense: there are at least processes for enabling expression. The piece allows you to remix, share, and browse stories made from its reasonably large (about 35,000 word) content library, and a lot of the affordances offered within are tools for more easily expressing yourself. Like cut-up fiction or magnetic fridge poetry, it’s a curiously restricted form of expression, but capable of surprising results. As with the editor in the 2004 film The Final Cut, you’re given a life and asked to make a story out of it: the possibility space is larger than it might seem at first. Your tools are an ability to explore how a certain place changes through time (an idea found in, among many other places, the fascinating 1989 comic Here), virtual scissors and glue, and (crucially) the ability to easily share your creations with others and browse what others have made. My goal in part is for people to not just explore 18 Cadence but also express something with it: a moment they particularly liked, an interesting juxtaposition they noticed, the story of a secret or a downfall or a quiet hero, even stories or poems that cleverly repurpose the existing text to make new things.
I’m fascinated by open-world games like Minecraft or Skyrim that let users create their own narratives out of emergent behaviors, but I feel like isolating and experimenting with that process of narrative-building is an interesting step towards building something more computationally sophisticated around it. How do we make narratives that are meaningful to ourselves, and how could intelligent digital storytellers help us in that process? 18 Cadence won’t win any AI awards but I think it points in an interesting direction for future exploration, and that, I hope, is cool too.
