Puzzle Game Design

Posted by (twitter: @blubberquark)
September 9th, 2014 3:40 am

I have seen many bad puzzle games in this LD. I have encountered the same problem so often during rating puzzle games that I felt I should write one long explanation instead of many short comments.

With a literal puzzle, you can see what pieces fit together, you know what picture you want to get in the end, and you can imagine what the result would look like if you connected two pieces together. You have a clear mental model of how puzzles work.

In a puzzle game, the player spends a lot of time thinking, planning ahead. In order to plan, they need to know

  • what you can do,
  • what the consequences of these actions would be and
  • where you want to go.

If there is only ever one possible action to take, there is no need for planning. If there are many possible actions, and all but one of them fail, the player can solve the puzzle by trying every action at every step of the way. In these cases puzzle games degenerate into very boring action games.

If the player does not know what paths of action are available to them or what consequences these actions lead to, there is nothing to think about.The player can only solve the puzzle by brute force. This is slow and does not make the player feel particularly smart. This is the difference between a riddle and a puzzle: In a riddle, we can only guess the consequences of actions. We arrive at the answer to a riddle through lateral thinking. Riddles often rely on background knowledge from outside the game. Stopping to think is just as important for a riddle as it for a puzzle. Solving riddles by brute force is just as unsatisfying.

If puzzle mechanics are well-designed, the player can learn what an action does after only a couple of examples. Games can facilitate this process by including sensory cues of the changes in game state, like floating damage numbers in a JRPG. Even if the player can inspect the game state before and after an action, it is better to make it explicit. Game mechanics that build on familiar domains like time, space or social interactions can be understood by analogy. This can be guided by the presentation of the game. Game actions that are analogous to actions in the real world can benefit from animations that call this to mind. For example, instead of letting a sprite vanish after using the “hammer” action, an animation of the hammer hitting the sprite and the sprite shattering into pieces makes the nature of the action more apparent.

Games like “Zelda”, “Myst” or “Antichamber” incorporate two-phase riddle/puzzles that turn learning of puzzles into a game mechanic. They often present a new and unfamiliar puzzle mechanic, which the player must then figure out. Riddle/puzzles do not dwell on the same kind puzzle very long. The main difficulty of these games does not lie in problem solving or lateral thinking, but in learning new kinds of puzzles. The riddles are embedded in a larger game world and give it a feeling of depth and mystery.

TL;DR: The cognitive skills involved in puzzles and riddles are problem solving and lateral thinking, respectively. The core aesthetic of puzzles is challenge. The aesthetic evoked by riddle/puzzles is discovery. Good puzzle games make the player think and let them plan ahead. The player must be able to construct an accurate mental model of the game mechanics for this to work.

Now a positive example: The game “Faced” by Sheepolution does everything right.

  • It tells you clearly what the goal state is.
  • It teaches you what you can do how how that affects the game world.
  • It is based on pushing and pulling. Players are familiar with this from the real world and can reason about the game state using naive physics. The game reinforces this familiarity through animations.
  • There are often multiple ways to solve a problem. There is no single correct action to take.
  • The player can roughly measure progress by the number of smiling faces.
  • There is not simple rule that leads to victory. The player must plan ahead.

I would like to hear your opinion on puzzle design in the comment section below.

Also please play my game Gunship Diplomacy, it does not have any puzzles.


6 Responses to “Puzzle Game Design”

  1. SteveSalmond says:

    Nice! Thanks for taking the time to write this up. I’ll have to refer back to your post for the next LD! You seem like someone who might have some good game design links or book recommendations.. (hint hint) ^^

    • rfgpfeiffer says:

      My background is in cognitive science, so I can talk about learning, analogies, problem solving and mental models. All of these are important to puzzle games. I can not really recommend any “game design” books apart from “A Theory of Fun” by Raph Koster and “Persuasive Games” by Ian Bogost, because I have very little formal training in game design.

      The theoretical framework I used for setting games like “Faced” or “Sokoban” apart from games like “Myst” is: The Mechanics-Dynamcs-Aesthetics approach https://en.wikipedia.org/wiki/MDA_framework

      If you are interested in the cognitive science part:

      Human problem solving was investigated by Alan Newell & Herbert Simon. This is an overview of some of their work http://www.cog.brown.edu/courses/cg195/pdf_files/fall07/Simon%20and%20Newell%20%281971%29.pdf

      In my opinion, behaviourist theories of learning do not apply well to strategy or puzzle games (Maybe better to action games). Constructivist learning seems to me a better approach. How we actually construct mental models from exposure to examples is still subject to research. We can sometimes carry over our real-world models into games. This becomes much easier if the games resemble real-world concepts on a surface level.

      In first-person games or puzzle platformers there are conventions that players know, so they already have a mental model of much of the game. In puzzle games the player must often construct the model from scratch, so you need to make it more explicit which conventions from the real world can be carried over.

      Would you like some links on the cognitive science of analogy making?

      • SteveSalmond says:

        Sorry for the delay in responding – I’ve only just figured out how to monitor comment responses on the LD site.

        Interesting – I can see how cognitive science could be a great tool in this area. Despite being a programmer by trade, it would never have occurred to me to employ formal analysis techniques to try to understand and/or create gameplay. Duh!

        Thanks very much for those links, I’ll certainly be using them to kick off some background research. Please do post more on analogy making if you feel so inclined.

        Cheers :)

  2. Lars-Kristian says:

    I totally agree with you and I try to incorporate some of the things into my own games. Check out my submission. http://www.ludumdare.com/compo/ludum-dare-30/?action=preview&uid=42086 I highly recommend the post-compo version.

Leave a Reply

You must be logged in to post a comment.

[cache: storing page]