Limbo level designer Jeppe Carlsen considers the player his worst enemy, but also his best friend. Through simplicity of control, but also by frequently leading players to their untimely demise, the 9-person Danish team at Playdead created a challenging XBLA title with over 300,000 copies sold thus far.
Carlsen began as a game programmer, but became more design-oriented over time, though he had no design background. The game's design was set by the director in the general sense, with a detailed video of the mood and setting, alongside the idea that the game should only use directional control and two buttons (jump and grab), and should never have any tutorial text.
They ran into problems from the beginning. The team had a scenario where the player needed to get up a ledge that was too high to jump onto, and there was a boat to pull, but since there was no tutorial, players had no idea what to do. Nobody knew to grab the boat, because the character had just ridden to this scene on it. So they made a new scenario, and forced the player to wind up near a pullable object, and put the character's arms out when he was nearby. They stick you there until you learn.
"You could call it a learning-by-dying game," he says. "We wanted the player to die a lot in this game, I think it really works with the theme."
Carlsen's level design methodology tends to center around leading the player to believe one thing, then pulling the rug out under them. When you're being so difficult to the player, "It's important that you also treat him nicely," he says. "You don't ever really get penalized for dying, it's just how the game plays."
Puzzles have no replay value, so the fun is in finding them out, Carlsen adds. In that case it's important to make the player learn from their deaths, and make them look entertaining.
He also alters level designs in order to make the player feel something different. "Earlier in the design (of a certain puzzle in which a spider knocks down a bear trap that's vital for continuing) we had the trap in the scene with the spider," he said, but this caused players to focus on the trap, not the spider, so they moved it away to another screen. He wanted them to feel helpless, and to use that space to experiment. "I think it makes the scene work a whole lot better, giving the player this slight frustration on not knowing what to do," he said. But they based cues on sound, you can hear when the trap falls, and so forth.
The rules set out at the beginning of the project "didn't make it particularly easy" to design the puzzles, he admits. Thinking of how to lay out and design the simple elements, and you may want to try to predict what players can do. Present one element so that they have to do it, like a switch that they will most likely push by running past. Then trick them to make them think some solution is the answer. This will force the player to really stop and think about what they're doing.
First, he says, think of the player as your worst enemy, and then create the most devious puzzle possible. But then from there, try to work with the player as your friend, so that you can give them the right clues. Start with tough stuff, then scale back.
Carlsen showed a number of early puzzle designs which were far too complex. They would see what worked in these puzzles, and take that one element, but through this process he admits that "we had to throw out a lot of work all the time."
"It's very important that the correct solution is fairly easy to execute," he said. If the player can't make the correct approach work, after a few times, they will stop trying that correct idea. "It takes the player a long time to go back to an idea that he had previously discarded as not being possible," he says. So, the wrong approaches should clearly present that they are wrong. If they can get close with a wrong approach, they can get very attached to trying that idea.
Limbo is linear, so you can get really stuck, and there's no hint system. "That's not really a fortunate situation to put the player in, but we didn't want to have any in-game menus or anything."
"The biggest rewards come from finally finding the solution to a puzzle you had to struggle with for some time," he concluded. "Getting that balance right is really important in making a puzzle game that's entertaining to play."