“Intelligent” Design

The other day I fired up Assassin’s Creed II to run around Renaissance Florence and, had it not been for a mandatory update I couldn’t access from my dorm, to finally try out the “Bonfire of the Vanities” DLC I picked up this summer but never played. When the latter failed, I settled for the former, and for a half-hour or so I enjoyed the general freedom of exploration, the smoothness of the combat, and the details of city life.

But around the thirty minute mark the enthusiasm began to wane. Having already finished the story, discovered “The Truth,” gathered all the feathers, and fully furnished Monteriggioni, there wasn’t much left in Florence. The bards, strumming merrily on their mandolins and hoping for a few florins; the shopkeepers, defending their extortionist rates; the heralds, denouncing my atrocious behavior — at first, it was invigorating, but it soon became insipid.

A few months ago Nathan Riley touched on why games as ostensibly immersive as Assassin’s Creed eventually fall flat: their worlds, while sprawling, are dead. Seemingly thriving cities are stripped of their facades by discerning players to reveal a complicated clockwork, but clockwork nonetheless. We know precisely what the “people” of Florence will say if we kill someone in front of them. Eventually, we have memorized the entire list of merchant lines by heart. I’m inclined to agree with Michael Ubaldi’s response: games must move in a more improvisational direction. And while it’s impossible to tell precisely what that fruit will look like, I think the seeds have already been planted, if in radically different places.

Exhibit A: Valve

Valve’s AI Director is impressive in its own right. Constant analysis of player behavior results in a slew of alterations to the game itself, ensuring that no run-through is the same as another. Are you barely surviving? Here’s a beeline to the safe house. Are you the second coming of Bruce Campbell? Good luck making it through the cemetery. It isn’t just a matter of easy vs. expert; enemy types, quantity, mood, placement — not to mention the physical restructuring of environmental components — are determined as a basis of real-time evaluation of your performance.

Exhibit B: Dungeons & Dragons

Similarly, consider the position of the Dungeon Master in a good old game of tabletop D&D. Armed with a thorough understanding of the rules and how to bend them, and charged with the task of actively crafting and adjusting the campaign in relation to your relative ingenuity (or incompetence), the Dungeon Master (DM) is the director of an improvisational drama. The enemies you encounter behave realistically and cleverly, more like actual opponents than the product of mathematical formulas. And yet, if your party is in danger of a game over, the DM can fudge the numbers. The enemy makes a mistake. He misses where he might have slain. If the DM behaves well, we don’t see the game as inconsistent; instead, we see the enemy as flawed; realistic.

What makes an encounter in D&D engaging is the knowledge that no one — not even the most meticulous of DMs — knew for sure how the fight was going to play out when it started. The thrilling “wait, guys, I think we can try this” works because the rules allow it to; instead of an invisible barrier or a faded menu option, we do as we please, and the game adjusts accordingly.

Exhibit C: NATAL

Sure, the infamous “Milo” demonstration of pre-christened Kinect may have been staged, but its potential (some of which has already been realized) gives great insight into where this spontaneity may enter gaming. Moreover, in light of impressive breakthroughs in language research spurred by Noam Chomsky, the concept of a self-teaching AI is within our grasp. Like the humans they emulate, computers should be able to apply an innate set of grammatical rules and apply them to unknown sentences. The incredulity of the Venetian bard resides in the limited range of interaction you can have: ignore him, pay him, shove him, kill him. But what if you could speak to him? And what if he could speak to those around him?

Of course, there are necessary limits. There are some things a DM won’t let you do — just because you think of saying “Petrificus Totalus” doesn’t mean an imp will suddenly become paralyzed. Your behavior is still guided by what makes sense and by what is within your character’s power at a given point in the game. If you can technically do it, and if doing it could theoretically make sense, then the DM will let you do it, and if that messes with the DM’s plans, so much the better for the game.

An artificial intelligence director with control over the game world would be able to understand what you’re trying to do and why you ought to be able to do it.So let’s integrate. An artificial intelligence director with control over all of the agents within the world and, genre-permitting, even the makeup of the world itself, would be able to respond to the most improbable of player behaviors (while still appropriate) because it can listen to your interaction with the world and understand what you’re trying to do and why you ought to be able to do it. In a game like Assassin’s Creed: Brotherhood, this could translate into a sort of group huddle, in which you tell your friend your ideas about attacking such-and-such a target, and he gives you a valid assessment of how viable your plans are. Meanwhile, if you have been too easily surmounting challenges lately, perhaps the director uses its ability to eavesdrop to help fortify a particular entrance to the enemy’s fortress that it knows you’re going to try to exploit.

With the prevalence of internet connectivity, the game’s AI could be synced with a central brain, thereby greatly accelerating the director’s language acquisition processes and adjusting all games based on common player strategies. In Left 4 Dead terms, this means the game learning about that one closet everyone hides in and making sure that whenever players are trying to exploit the closet, a Spitter shows up.

I appreciate the difficulties involved with actually implementing what I’m suggesting here. But if Kinect ever performs on the level of that Milo demonstration, I think we’ll be a lot closer to games that are truly immersive. Just think; in the world of real advertising, you can’t just use the same slogans forever. A tapestry merchant who knows you’ve already heard his pitch improvising something new based on what’s worked elsewhere? How about a bard who legitimately knows a song you’ve never heard before? NPCs that relate messages from other players, rather than the faceless couriers of old?

The potential, endless. My spine, tingling.