Gamasutra: The Art & Business of Making Gamesspacer
Game Design Theory Applied: the puzzle of designing a puzzle game
Printer-Friendly VersionPrinter-Friendly Version
View All     RSS
April 16, 2014
arrowPress Releases
April 16, 2014
PR Newswire
View All
View All     Submit Event

If you enjoy reading this site, you might also want to check out these UBM TechWeb sites:

Game Design Theory Applied: the puzzle of designing a puzzle game
by Toni Sala on 01/02/14 02:46:00 am   Featured Blogs

The following blog post, unless otherwise noted, was written by a member of Gamasutra’s community.
The thoughts and opinions expressed are those of the writer and not Gamasutra or its parent company.


Today’s post is the fourth in the series “Game Design Theory Applied”. You have the previous ones here:

Today I would like to talk about the tricky concepts around puzzles and puzzle-based games like New SokobanPuzzles are often considered to not be games at all. This means that designing a video-game entirely based on puzzles implies some issues that need to be addressed to minimize the inherent problems that puzzles have and maximize their benefits. We will see it applied to New Sokoban, a puzzle-solving game.

Game Theory Applied: puzzle games


Puzzles are not games… really?

Well, indeed it’s a tough question. Jesse Shell wrote an entire chapter about this topic in The Art of Game Design and didn’t convinced me completely :p However, intuitively, we feel like puzzles are slightly different from games or lack something that puts them apart:

  • A puzzle always has a dominant strategy. And one of the main objectives of a game designer is to avoid dominant strategies in his games. Paradoxically, when designing a puzzle, you are deliberately creating a dominant strategy: the solution. In the particular case of puzzle games, the goal (and the whole fun) is to find the dominant strategy. The problem is that once you get it, the puzzle ceases to be fun.
  • Puzzles often suffer from the enigma syndrome. I mean, sometimes, to solve a puzzle you need to make a perceptual shift. Can you arrange six matchsticks so they form four equilateral triangles? Have your ever faced a problem like this one? This kind of puzzles often need us to make a perceptual shift to be solved. And it is a double-edged sword: if you make the shift, you solve the puzzle and the satisfaction is huge. However, if you don’t, the frustration and shame is also noticeable. The problem is that it is a matter of inspiration. No game mechanics are involved.
  • Puzzles often lack of Triangularity. Triangularity is a very important concept in games. Triangularity is present in a game when an easy and low-risk strategy leads to low rewards while taking the difficult and risky path leads the player to generous rewards. The more triangular decisions there are in a game, the more interesting and fun it is. The problem is that in puzzles there is usually one optimal solution. So, triangularity is very difficult to introduce.
  • Puzzles are not replayable. This is the key problem. All the other problems lead to this one. And this is the main reason for why puzzles are so often not considered games.

Following in this article I’m going to discuss how I tried to minimize the negative effects off all the above problems in New Sokoban. But, first, why indeed are we interested on designing puzzle-based games? What’s the point on bothering to design this kind of games? Because we, as humans, love them. Puzzles are essentially problem solving tasks. And these kind of tasks are appealing to our human brains.

Game Design Theory Applied to New Sokoban: creating a game of puzzles

From my point of view, the key thing is to realize that, actually, you are not designing a puzzle-game but you are designing a game made of a number of puzzle instances. A well designed and balanced collection of puzzles may lead to a great game.

It is very difficult to avoid the problems described above (at least for me), but I applied some simple tricks to try to minimize them. For example, to minimize the enigma syndrome I offered lots of puzzle instances to the player with a well designed and balanced difficulty curve.

As you keep solving puzzles in increasing difficulty, your mind gets used to that and the perceptual shift effect is minimized. Moreover, in some sense, I positivate the perceptual shift problem when the player completes a world and advances to the next: the first puzzles of the new world always require a perceptual shift (not a very difficult one, though). However, it is for surprising purposes. Surprise is always a desired concept in games.

Moreover, the player can always skip any puzzle just using the button for this purpose on the pause screen. Maybe later he will be able to make the perceptual shift needed to solve that puzzle.

To achieve the “illusion” of Triangularity in New Sokoban I introduced the bonus system. When you have played for a while, you know that if you solve the puzzle without using the undo or reset options you get a higher score. Moreover, the less time you spend on solving a puzzle, the higher the score you get.

So, based on those variables, some players take the trial-and-error strategy. They focus on solving the puzzle and forget about the bonuses. They perform a lot of (non-sense) moves, go back, and try again. Other players, try to solve the puzzle in their mind as fast as they can and afterwards they execute the movements that lead to that solution.

This is not exactly Triangularity, but indeed it is an emergent mechanic based on the trade of between reward value and the risk of performing certain non-correct movements.

Minimizing the dominant strategy effect is a tricky task. That’s because, as we said previously, a puzzle is actually all about finding the dominant strategy. To deal with it I introduced the stars system and the under par system.

This allows me to reward the player for achieving a sub-optimal solution and, at the same time, encourage him to find the optimal one. Actually, this doesn’t solve the problem, it only moves it a little further: when the player finds the optimal solution, that puzzle ceases to be fun.

However, I really don’t inform the player about the optimal solution (the maximum possible under par). The player knows that every single puzzle may be solved with 3 stars but he doesn’t know which is the maximum under par for every puzzle. So, in theory, the player could come back to try to find the optimal solution for the eternity. The truth is that in some cases this leads to frustration… and, as I said in some other articles, that’s the worst word for a game designer…

Finally, about the replayability. Actually, addressing the previous issues improves the replayability of New Sokoban. However, I plan to introduce new features to the game that will probably improve a little bit more the replayability. In a few words, I will introduce sub-objectives in every puzzle: achievements if you stick to some constrains, McGuffin objects, etc.

In fact, this kind of features will allow me to present different instances of puzzles in the same board configuration. In other words, every puzzle will have more than one optimal solution, one for every different goal.


As you can see, designing a puzzle-based game is a nice and challenging task. I tried to face it from the optimistic perspective: “I know that a lot of people like puzzle-solving tasks. Let’s see if I can design a game full of that!!

New Sokoban is getting incredibly good reviews around the world: 4.75 out of 5! Moreover, I know a lot of people that hates puzzles but are enjoying New Sokoban. So, maybe New Sokoban is indeed a fun game made of challenging puzzles. 

Toni Sala  ":^]

Related Jobs

SOAR Inc. — Mountain View, California, United States

Game Designer/Narrative Writer
Giant Sparrow
Giant Sparrow — Santa Monica, California, United States

Game Designer
Nintendo of America Inc.
Nintendo of America Inc. — Redmond, Washington, United States

Associate Software Support Engineer
Nintendo of America Inc.
Nintendo of America Inc. — Redmond, Washington, United States

Software Support Engineer


Darren Tomlyn
profile image
There's much more to puzzles than this, but that means understanding what they are - which is currently as much of a problem as it is for games.

No, puzzles are not games at all - they're a completely different activity, (or a thing that happens within and as part of an activity), that just happens to be compatible with games in one particular way - (a race to complete a puzzle) - that can only be referred to as the true 'puzzle game'.

Unfortunately, your understanding of puzzles is too limited, which is unsurprising, since confusing what a puzzle is with how it's applied is the most common problem they have.

This is because people are not recognising such activities as and by what matters - the BASIC things that happen within the activities, on behalf of those taking part - (which is itself a symptom of a far bigger problem with our understanding of language (and possible communication)).

This has caused a lot of problems with various words people try and use to describe all types of such things that happen, often regardless of the activity - such as action, interaction, choice/option/decision, emergence/emergent etc., since they do not help to understand the difference in definitions, only applications (which are always subjective).

Although my current blog kinda deals with this, it's no longer fully consistent and accurate, given what I know now and understand, though the solution I have for this problem has no issues.

I still recommend you read it, though the first two parts (1A&B) are problematic:

Luis Guimaraes
profile image
Assuming Fantastic Contraption and Scribblenauts classify as Puzzles, they don't suffer from any of the problems mentioned in the article.

Bart Stewart
profile image are you selling a game called New Sokoban? ;)

I have no problem saying "puzzles aren't games" to game developers, because they know that games are just one form of play... and play is what actually matters. (I'll still call puzzles "games" to non-game developers, though.)

The model I use to try to understand fun says that game-play is about gaining extrinsic rewards (loot, money, status) by being good at following rules. Puzzle-play is about satisfying the desire for the intrinsic reward of feeling clever at having perceived the pattern that solves a puzzle. You can make something that combines those two kinds of play, but they are different styles of fun and need to be understood as such. Building a pure puzzle and expecting people who prefer rules-based games to appreciate it is just asking to be disappointed.

One suggestion is to change from thinking in terms of "the solution" and design puzzles to reward "a solution." It sounds like you've tried that. Another great example of this is SpaceChem, which actually shows players where their solution falls compared to others. This is brilliant -- it lets game-players who want to win do so by proceeding after devising any solution; it's highly repayable by the gamers who enjoy trying to optimize their solutions; and it's still a puzzle so it rewards perceptiveness. And because SpaceChem has two different ways of measuring solutions (number of components used and number of cycles processed), there is no single "best" solution.

I would be careful about using time limits to increase what you called triangularity. That's fine for rules-based or sensation-based play, but arbitrarily cutting off thinking time can be extremely frustrating for players who are trying to perceive the patterns of a puzzle. Offering a bonus for finding more elegant solutions is OK, though.

Michael Ball
profile image
If puzzles aren't replayable, why are movies rewatchable and books rereadable? Serious question.