Kung Fu Legends
Kung Fu Legends is a roleplaying video game currently under development. You play a Kung Fu Master killed by a rival dojo. During the attack on the dojo, your young students escape. You mentor one of the students from the spirit realm, hopefully guiding them to become the next Kung Fu Master. Will you encourage him to seek bloody revenge? Explore the world? Learn hidden styles of kung fu? Take on the Emperor himself?
Rationale
There are a lot of Role Playing Games out there at the moment: running the gamut from Diablo to Oblivion, to the mighty Dungeons and Dragons, to Final Fantasy, Nethack and World of Warcraft. Many of them are built on similar mechanics: you control one or more characters with explicit statistics on their abilities, inventories and opponents. As much as I love these kinds of games, I want to play something different. I'm sick of micromanaging a squad and spending time determining if Pants of Frosty Fire are better than Pants of Flaming Ice when I could be off experiencing a story. Kung Fu Legends in this way is more about macromanagement: strategies, personal development and plots. You don't need to manage inventories. You never see objects purely in terms of their combat efficiencies. I want Kung Fu Legends to tell a story.
This is not to say that the world isn't simulated. We have powerful computers nowadays, so there's no reason that RPG simulations need to be reduced to the same rules as a table-top game. Moreover, there is no compelling reason why the player needs to see this machinery. In a Role Playing Game, they should see the effects on the characters and narrative, rather than worry about a +1 or +2 combat modifier. In Kung Fu Legends, you don't get access to the underlying mechanics, nor would you want to. I want Kung Fu fights to be a simulation of flying fists and swift footwork, rather than roll versus a success.
Another point of difference I want to make is reconsidering determinism and replayability. We have a lot of experience with games that operate randomly. Given a fight in, say, D&D, the outcome of a fight is both a product of your choices and how those choices are modulated by dice-rolls. Suppose the game world was deterministic. Given the same inputs, the same things would happen. But if something changes, then the complexity of the simulation gives rise to chaos — a small change in plan could mean quite an eventual difference and it's solely due to your choices, not in what came up in a random number generator. This capability allows you to play through the same game differently and explore what your choices mean on a micro- and macro-scale.
There is some exciting new work being done on AI Game Narrative. For example, in Left 4 Dead 2, the AI Director will modify maps, music, zombie placement and cue exciting events to make the game more fun for all involved. Spelunky (like its grandfathers, the rogue-likes) generates sensible levels on-the-fly for you to run through. I am working on a system that generates narrative for Kung Fu Legends. Since the game knows the entire world model it can explain the world model to the player. It can do this through text, which is easily extensible and doesn't have to worry about some of the details you need for a graphical simulation (for example, you can say a character walks across a room quite easily, but making sure a digital avatar doesn't bump into objects, steps correctly, turns correctly, stops correctly and does it all in a convincing manner... that's hard). Of course a computer can't be expected to produce life-changing prose, but it can write something servicable for a game. And with the Storyteller AI working in cahoots with the character AIs, they can help create a game world that is both complex and interesting for the player.
These ideas are ambitious, but I think I have a good way of achieving modest versions of these goals. Hopefully I can provide something new to the RPG scene.