I released a silly little game called Crowd of Three over on my other blog. It's a game vaguely related to strip poker, but don't go expecting porn: that's not why I built it.
This post will explain why I built it.
I want to build social AIs. The problem is, a social AI needs a world to inhabit, or it has no "hooks" with which to do meaningful socializing. Most simplistic worlds either (A) don't have the depth of emotion I need or (B) don't have the breadth of emotion I need.
For example, a game about an unhappy couple. It doesn't appeal to me. There's not enough complexity to the socializing unless you put that unhappy couple in a world full of content, which quickly takes it from "simple" to "800+ megs of poor line art and dialogue trees". Not to mention any names, of course.
So, a few weeks back, I was thinking:
Me: "I need a really simple world with strong built-in emotional hooks. Preferably a game. What kind of game gives you a strong emotional feeling as you play?"
Me: "Hmmmm... well, any game where you have a big stake... like poker: to the death!"
Me: "No, it needs to have a broader range of emotions. Friendship, rivalry, shyness, anger, gloating... what's a simple situation that has all of those and makes them change over time?"
Me: "Hrm... toughie. I give up. What?"
Me: "I was hoping you'd know! Why do I have to think of everything?"
The first answer I stumbled across was Strip Poker. There's not a whole lot of other games which result in as wide a variety of behaviors, and it's fairly simple.
Of course, strip poker wasn't enough. I took the parts of strip poker that aided social interactions and punched them up a notch. I took the parts of strip poker that reduced interactions and killed them. What I ended up with is a very simple game of die-rolling where you can give chips to other players. In exchange for gratitude or clothing.
Regardless of the fact that strip poker games have always been strictly pornographic, I really couldn't think of any other game which could serve the same purpose. So... strip poker it was!
As it turns out, the social AI I implemented was too simplistic - or too clear, depending on how you look at it. This is partially because the algorithms I used were simplistic, and partially because the game has a very restricted "team play" dynamic, making it hard to express your like and dislike in the play environment: it's strictly social, which isn't good enough.
I think my next game will be about a dream. No dialogue at all. :)