When friends ask me what tabletop role playing is all about, I tend to describe it as collaborative storytelling. Sure, as a game master, I should be keeping an eye on where my players' journey is taking them, what events and challenges will come on the way, and how the story world and society works. But it's not my job to tell the story - at least not on my own. That's for the players to do. I've too often fallen into the trap of overplanning how a story unfolds, coming up with intricate challenges with precise pre-determined resolutions for the players to discover or figure out. But I've learned this generally leads either to frustration and getting stuck, or an unenthusiastic "okay, well that's over". The story events that really liven up the atmosphere and bring the energy and excitement are when I don't already know what's going to happen - I just present a situation and see where they take it. This sense of true player agency, using your own initiative and creativity to drive a story forward in the way you decide - that's where we get our RPG kicks.
And it's something we look for in computer gaming too, evident in how story-rich games have evolved over the years. It was once the norm for story environments in video games to be organized into nicely scoped scenes with rigid transitions between them - remember the Monkey Island series? Aside from providing an easy solution to technical challenges like the memory demands of large game environments, it also gave designers more control over how you experience the story - story events are tied to scenes in the game, and you experience these events in a controllable sequence as you hop from scene to scene through carefully curated transitions. But then came open-world gaming: a paradigm that's moved the needle on player freedom and immersion. Open World Gaming gives players the ability to roam freely around vast complex game environments, choosing for themselves what encounters to undertake and when, without being railroaded through a predetermined sequence of scene transitions.
The main draw of open-world games is about providing the player with autonomy—not so much the freedom to do anything they want in the game (which is nearly impossible with current computing technology), but the ability to choose how to approach the game and its challenges in the order and manner as the player desires while still constrained by gameplay rules.
At Mind Mage, we want to shift that immersion needle even further and achieve the "impossible" through an approach I call Open Narrative Gaming. Whereas Open World Gaming cedes control of player movement from the designers to the players, we want to hand over control of how the events play out. With Open World Gaming, you don't select where to go next, you just go. With open narrative gaming, you don't select what to do next, you just do. Today, even in the most sophisticated of open world games, the ways you interact with the environment are limited to the options presented in a context menu or action bar, and chatting with NPCs works by choosing dialog options designed by a writer.
We want to change the way we think of game narratives - writers build worlds, characters, quests, and events. They don't build resolutions to these events - that's the players' role. You want to fashion an improvised jet ski from the fire extinguisher you stole earlier? Sure. You want to try persuading the mayor his political rival is in league with your nemesis? You got it. In truly immersive narratives, the players are co-authors.
Screenshot of dialog options from Baldur's Gate 3 © 2023 Larian Studios. Used under fair dealing for the purposes of criticism and review.
It's a lofty goal: to facilitate this co-author relationship between players and designer, we need to push the boundaries of generative AI technology, creating an AI storyteller that can truly understand both a designer's vision, and how players' actions relate to their environment and the overarching story line. Today's Generative AI technology is excellent at coming up with superficially coherent dialog responses. But can it evaluate a player's actions - validate them, decide their efficacy, and determine the consequences in light of a carefully designed story and game world? That's where we have work to do. If we let players convey any action they want, how does the game know when they assert or presuppose something contrary to the designer's vision, or fundamentally undermine the challenge of an encounter?
Let's sketch out some of the kinds of hurdle we are solving here. Picture this: as a game designer, you put a really intriguing challenge before the players in which they'll have to come up with a creative way to escape a prison cell with various artifacts to inspire escape plans.
Player: I remember I took the key from the guard earlier - I take it out of my pocket and unlock the door.
Story teller: Slow down there. I'm afraid you didn't take any keys from the guard earlier.
The engaging game experience we want to give is one which rewards creativity, and creativity is born out of navigating constraints. I believe it was Orson Welles who said "the enemy of art is the absence of limitation." Players should be able to attempt anything they want, within the limitations of a setup - to feel like they've 'beaten' the constraints in front of them. If they can circumvent it by magically invoking a deus ex machina macguffin, there's no sense of achievement to be had. But how do we build AI capable of distinguishing between game-breaking presuppositions, and benign assumptions like that there's oxygen in the air, or that the walls are solid?
Another challenge to overcome is aligning conversational references to environmental features with the environment's true design.
Player: I notice a chandelier hanging above the bandit, and aim my bow at the flimsy rope connecting it to the ceiling
Story teller: I'm afraid your eyes deceive you: there's nothing above the bandit's head.
Player: Excuse me, could you tell me where the kitchen is on this ship?
NPC: Ah, the galley is downstairs in the aft of the boat.
Clearly, when environmental features are integral to the design of the events, the conversation must be faithful to that design. But what about when a player enquires about environmental features that weren't specified by the designer,, like the temperature of a key, or the texture of a door? With a truly open narrative experience, it's not practical to anticipate the infinite possibilities at design time. We need a system able to improvise, and to make a judgement as to whether the features it wants to describe would detrimentally influence the story. And furthermore, it needs to stay faithful to these new improvised details throughout the remainder of the story.
Join us here as we share more in our journey to build experiences giving complete immersive autonomy to players, without compromising the integrity or coherence of a well thought out story.