This is a spinoff of a comment on Emily Short’s post on conversation models, which got off-topic enough that I thought I should put it here. It’s something I’ve been thinking about some anyway.
Crayon Physics Deluxe is a game about invention, where you draw shapes on the screen and they come to life as objects with reasonable physics–a curve becomes a ramp, a block falls down, a hammer on a hinge swings. It’s full of emergent behavior, and every level surely has a solution that no one’s ever thought of before.
It also spends a lot of effort encouraging you to be awesome. There’s a splash screen at the beginning urging you to go for awesome solutions, and you can literally declare your own solutions to be awesome within the game–in fact, you can’t unlock the last levels without declaring a lot of solutions to be awesome.
Now who wouldn’t like to be awesome? Being awesome is awesome! So why does the game have to make us be awesome? Because the goal is always to get a ball to a star (or two), and that can be quite fiddly. After five or six times when the ball stops short because your ramp is just a little too bumpy, you begin to appreciate the plodding reliable solutions. I’d rather be sending rockets flying all over the place, crashing into things and setting off chain reactions, but too often that goes unrewarded. There’s a level with a rocket car–and the easiest solution to implement involves getting rid of it so it isn’t in your way! (There are other more elegant solutions that use the rocket car itself, but they take a lot of trial and error.)
The problem is that the awesome stuff is just a bit unpredictable–a scoop that just happens to golf the ball into the exact right spot, some ropes that slingshot the ball across the screen in a way I could never reproduce. But when it doesn’t come out right it’s unrewarding and often leaves the screen a total mess. What would make doing awesome stuff more fun would be for the near misses to lead to cool things even if they didn’t do exactly what you wanted. Drop the ball in a hole and another rocket shoots up and sends it somewhere; roll into a side path and find yourself somewhere where there’s something to do. Like that. And it may not be entirely compatible with a game where you’re always trying to get a ball to a star.
(This last bit was inspired by the discussion in Emily Short’s post about Framed, a game where you’re rearranging comic panels to change the narrative–but it’s always about helping someone sneak past the cops. BTW if you’re thinking “I wish there was something that vaguely resembled that but was a lot more interesting” check out the demo of Gorogoa.)
So the lesson is the Dwarf Fortress motto: Losing Is Fun. If you want to reward the player trying awesome but hard-to-predict stuff, you need to reward them for something even when they don’t quite succeed at doing what they want. Which means narrowly constrained goals aren’t the way to go. Maybe the goal could be something relatively easily obtained so the player can show off and experiment while they’re doing it, or maybe there isn’t a defined goal, or the player can pick new goals as things change. It’s my understanding that Minecraft does this–it gives you a goal of surviving the zombies but once you get the hang of it it’s simple enough that you can do whatever you want. Or have a mechanic that propels itself along and lets the player ornament it (or maybe derail it, if they want).
What we can’t do as game designers, if we want players to try cool unpredictable stuff, is punish them harshly when that doesn’t go just as they plan it. If doing something almost awesome means they have to start over, and doing something plodding means they make progress, plodding will win out every time.