Jeppe Carlson (Playdead): 4:30pm
(official GDC brief) (note: I was 5 minutes late)


puzzle designer discusses his process for designing much-lauded physics puzzles in recent indie game hit, Limbo.


Mostly interesting to hear about the tools they used. One of their programmers created an AI element which automatically playtests the level every time you compile it (run the level, and instantly see what most players would do in it). I think this was a huge insight into why the game turned out so well. Ingenius idea (to simulate the average player).


11 members on the team, took 3 years to develop the game … Jeppe’s main role was puzzle design, play testing, and difficulty adjustment … Seems to be saying the key to their success was the speed of prototyping (level editor)

Design principles:
Challenging puzzle solving.
Simple elements.
Very few elements.
Complex problems in simple environments.
No repetition.

Notes how uncharted 2 presents the illusion of a puzzle, but really you just have to find the mirrors and press a button to move on (No thinking required). In contrast, he talks about a Prince of Persia puzzle with tons of pools and doors, which left him overwhelmed – so he just trial-and-error’d it until he passed. The solution came through experimentation, Not thinking!

He learned : if you can’t boil your puzzle down to few elements, then it probably isn’t a good puzzle.
The main challenge was: lowering complexity
… Goes through chain puzzle in detail (Can’t reach chain unless it is swinging before you hit the trigger to move it). This puzzle only really has three elements…

Ideal play experience is:
– Experiment and fail.
– Learn by dying.
– Analyze the scenario and derive strategies.
– Ultimately solve the problem.

… mentions that each area is entered with no preconceived notions of what you will be doing… this approach leads to enjoying a much more rewarding feeling when you solve it.

Talks about the importance of trying to predict how player will approach the puzzle.
Chain Puzzle Example: Player enters left to right. They will almost certainly presses the button first. Chain moves slow, teasing. But you think maybe speed is solution. … (So they kind of require you to do the opposite of what you first thought.) … By showing the edge of death pad, you want to swing to see if you can reach. This leads to hint (swinging is the solution, just not at the end of the puzzle).

In the designer’s mindset, the player is the enemy, so he come up with devious tricks to mess with them. … But the player is also a friend, so he designs accessible environments to help them.

Solution must ultimately be easy to pull off. This game is about frustrating the player. He shouldn’t think it impossible. Can’t let him be distracted by wondering if he needs more skill (better jumping reflexes).

Attempting the wrong solution must communicate that they are wrong.

(To keep them from hammering into it over and over again). And they should be able to learn something from their death(s).

In their level editor, the boy primitive has ai when you start playing the level. He explores limits of his space. Hangs off edges of platforms. Useful for physics based games… Their editor also let’s them fiddle with values (weight of box) while the game is running. (they didn’t think to include a feature to store these fiddling values though. Heh.) … “A very talented programmer worked on the kid primitive’s AI for three years.”


Q: Someone asks if having so many check points in the game encouraged brute force solution.
A: Jeppe claims they kept the the scenario’s simple enough that you couldn’t draw the wrong connection between them (.? Errr… Not sure that was the question’s point). Jeppe notes they definitely wanted to avoid brute force solutions.

A: They play test internally first. A good barrier, because co-workers know Jeppe’s puzzle-design mindset well. Eventually, they invite gamers in off the street. They don’t ask tester if they like it, or like the graphics. No filling out forms or asking questions. They just feed communication back to Jeppe.
“Did they get stuck in frustration loop: yes or no” is basically the only thing Jeppe notes. (to address somehow in future puzzle tweaks)

A: The whole game only exists because of one artist’s initial concept video. He sat at home working on art until he was sure he had a cool aesthetic. Then Jeppe knew the game would need to have jumps and clinging mechanics – from the video.
(w: interesting contrast to my monopoly design idea. that there is no place for an artist’s uncompromising voice in game design…)

Print Friendly, PDF & Email

Leave a reply

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>