I Am a Dice: Using Humans as Procedural Content Generators

I'm extremely interested in procedural content generation, but one of my peeves with PCG as a whole is that it tends to be so basic, so single-layered, and fundamentally random. I'm continually thinking of ways to make the generated content more meaningful and rich.

One thing that has come to me lately is the idea of humans being the 'random' element of a procedural system. As I see it, procedural content is made from two things: Rules, and generators. Usually all the interestingness, the gameplay, comes from the rules. The generator is just random noise, or Perlin noise, or a normal distribution. So what if we use humans as the generators?

Now, humans producing content within rules is nothing special: that's what making games is! The trick, I think, is in making their input simple and easy, and building it into the gameplay, so they don't realize they are producing content. Let the rules make it interesting, like in a procedural system. So what does that give us?

Learn from the Board Games

The game that inspired this thought was Taluva. It's a board game where players take turn laying and stacking tiles, and then placing tokens on those tiles. Where it's interesting is that players must place a tile each turn, there are very specific rules for where a tiles can be placed, and the tiles them selves have no value towards winning the game. At the beginning of her turn, a player places a tile in a way that benefits them the best on their next placement or screws over an opponent, and then places a token. After this, the tile 'fades into the board'. This new tile will have an effect on the placement of future tiles and future pieces. But there is never (well, rarely) a feeling of "Ugh! Graham's tile is in my way!" Players don't have any sense of ownership over the tiles. They simply become part of the 'generated terrain'.

Another game that many of you might have played that has a similar is Carcassonne. Again, each tile is very significant and gameplay focused at the time of placement, but quickly becomes just another part of the landscape, keeping the game fresh and unique, just like procedural content is supposed to.

No computer games have come to mind that really exploit this kind of feature. Obviously in many games, strategy games in particular, the actions of the players in the game 'change the landscape' for each other, but it's not in that anonymous, neutral way that procedural content allows.

Theory Without Practice...

So here's a few quick ideas I had for how this could apply to games.

First, a Diabo-style dungeon crawler. A kind of game that relies heavily on PCG for item selection and loot. So imagine this: Every item starts out 'basic'. It's as crummy and useless as possible. But then, through the players direct or indirect actions, they modify and upgrade the weapons. A Copper Dagger becomes a Copper Dagger of Stabbing through repeated use of the stab action. The embedding of a fire rune turns it into a Flaming Copper Dagger of Stabbing. But alas, the time comes when they find a basic Steel Dagger and decide it's time to move on. They sell or drop their old dagger on the ground, and it's now fair game for another player to pick up.

This example might not be immediately useful; a bunch of new systems with no effect on the second player's gameplay. But I'm sure you can start imagining a lot of neat ways to take this to enrich the game and build community.

Another example that jumped to mind was some kind of RTS where after each online skirmish, the map is saved. Each time players start a new skirmish, the map is filled with the remnants of a previous, anonymous battle. Structures and walls make various ready-made bases. Resources have been extracted and redistributed around the map. The remnants of another player's strategy linger in the air. I think this could be cool.

So, any games you can think of where the gameplay of one player is the ambient content of another? Where the actions in the early game shape the late game, but have minimal long-term strategic value for the player? Or any other cool ideas where this could be applied?

Comments

[Jordan]

Just thinking about this, but wouldn't Spore fall into this sort of gameplay? I get the feeling that you're not so much talking about player created content and more of player caused content so this might not be a good example. But it's all I could think of.

[Graham Jans]

That's interesting: Spore is a really obvious example, isn't it?

I also think that's a great term: "Player-caused content." I'm going to have to remember that one.

[Nels Anderson]

It's not quiet the same as content generation in Carcassonne, but Versus Mode in Left 4 Dead has some similar aspects. L4D makes dynamism fundamental even in Survival mode, but Versus mode takes this further, giving the special Infected the unpredictability only a human intelligence can provide. A team of coordinated Infected can give the Survivors an experience that's pretty distinct from what the AI alone can provide.

To me, at least, this feels different than other FPS games where the opposing team's goals and abilities are identical to your own (or at most, attack vs. defend). I've been thinking a bit how this might be utilized in a cooperative, or at least not directly competitive, fashion, more similar to what Carcassonne provides.

BTW, I came across blog your blog just yesterday. It's fantastic to see more Vancouver people blogging intelligent about games. Maybe I'll see you around at some local industry event.