© 2019 Spirit AI Ltd. All rights reserved.
Text-based games have been around since the 1960s, expanding and evolving into MUDs (Multi-User Dungeons) and other popular interactive fiction formats. These games captures the hearts and imaginations of users, and their mechanics can still be seen in games today.
I recently was invited to speak on behalf of Spirit AI at the Virtual Beings Summit in San Francisco, a unique group of innovators working on various facets of one of today’s most interesting challenges: creating lifelike digital characters and avatars. Maybe counterintuitively, I decided to use my time to focus less on the cutting-edge work we’re doing with Character Engineand more on some lessons from the past.
I believe we’re on the cusp of a huge revolution in how stories are told. The tech maturing today will open a whole new avenue for writers and storytellers to share their ideas. The thrust of my talk was my belief that the biggest challenge we’re facing right now in this space isn’t a technical one: it’s creating the quill, the toolset that will allow creators to leverage AI to help bring their visions to life.
In the rest of the talk I took a survey back through other moments in history when our tools for capturing and sharing ideas had to change. In the 7th century, for instance, St. Isidore of Seville lamented the fact that the sounds of music vanished and there was no way of writing them down. This seems incredible, today, that you could wonder whether it was even possible to record musical melodies: but in St. Isidore’s world there simply was no technology to do so, nor was it clear there even could be. While other cultures unknown to him had, centuries earlier, solved this problem in various ways, the European tradition had yet to catch up. It wasn’t until the 11th century that a Benedictine monk, Guido d’Arezzo, devised the core of what would evolve into modern staff notation, a system which allows anyone who learns it to accurately record and share music, and which has survived as a standard now for coming up on a thousand years.
Printed books went through a long period of experimentation and transformation after the invention of the printing press, when things that seem incredibly obvious now had to be worked out through trial and error. Page numbering, margins, bindings, tables of contents and indices, the sizes and shapes of books, the legibility of fonts: these and many other details had to be slowly and painstakingly evolved. Early bookmakers did this work of experimenting: they didn’t just settle for imperfect traditions carried forward from the days of hand-inscription. They figured out what worked in the new world of the printing press.
Even something as straightforward as the computer mouse signaled a shift not just in technological capabilities, but in how everyday people interacted with computers. Input for most users in the 1980s was becoming less and less about entering complex commands and more and more about choosing from a range of options that could fit on a screen. Here the right tool was something radically simpler, and that streamlining opened up personal computing to a new and much larger audience.
There’s a famous anecdote about how when Nietzsche first sat down at a typewriter, the first words he wrote on it were “Our writing tools are also working on our thoughts.” I can personally attest how painful it is to use the wrong tool for the job of creating interactive content. Whether you’re trying to write the middle of a branching conversation in a word processor, forcing you to try to remember the different possibilities that might come before or after which you can’t see onscreen; working on procedural text in a spreadsheet program that can’t even word wrap properly; or stuck in a proprietary tool that’s falling out of date, bloated with abandoned features, and prone to crashing: not only are these tools frustrating, they don’t help you think procedurally. You forget about the cool nonlinear affordances your system supports and start writing regular dialogue scenes. It starts seeming too difficult to do something truly interactive and you simplify for the sake of your own sanity. Your writing tools are working on your thoughts.
We’ve talked a lot on this blog about the ways we’re trying to build a better authoring tool for Character Engine. It’s a long, long road to build a genuinely new kind of tool, and we’re certainly not there yet. But I firmly believe the most successful platform for virtual beings won’t be the one with the flashiest tech or the smartest code. It’ll be the one with the best tools.
Dialog interactions are present in the majority of games developed in the past two decades. These interactions deepen immersion and give a greater since of investment in the story and game itself.
Most of these dialog interactions, however, only use basic branching. Dialog branches form a dialog tree of different explicit choices and results. In a basic tree the step from one choice to another is deterministic without consideration of any other outside elements.
All dialog that uses this approach has the same basic mechanic: select the correct options in order to get your desired result. While this is sufficient for many games, going beyond branching dialog can create a wealth of dialog mechanics that create compelling, complex, and varying dialog interactions and puzzles.