Among the many game genres, puzzle games may be one of the easiest categories for players to love and understand, and this category is often referred to as puzzle solving. In the previous "Independent Game Movie", many excellent explosive independent games are puzzle gameplay, such as "Fez", "Time and Space Illusion" and so on. However, the design of the puzzle game may not be so familiar to many people in their domestic counterparts.
In a previous GDC presentation, Brett Taylor, founder of independent game studio My Dog Zorro Studio, talked about the art and science of puzzle game design through the perspective of human cognition, and summarized 7 design experiences based on his team's research and development experience in the development of the first game and years of experience in puzzle game development.
Here's the full content of GameLook transcription:
Brett Taylor：
I'm Brett Taylor, I got my master's degree in cognitive science in Colombia and have been working on research and development for over 10 years. In 2015, I started my own studio, My Dog Zorro, and made a game called Linelight, so let's talk about Linelight.
Linelight was our studio's first game, and of course, it's a puzzle game, released on PS4, Steam, iOS, and Android in 2017, and between 2017 and 2018, it made about $100,000 after tax, so you can get a rough idea of what my annual salary is.
Its rating performance is also very good, with 90 points on the M station, and the iOS and Android platforms are also rated at 4.9 stars. If you're not familiar with it, watch a promotional video:
This time the content is mainly divided into four parts, first of all, the working memory, and then will mention some of the characteristics of the puzzle game, the more important part is from the "Linelight" research and development of the 7 experiences, of course, learned a lot of lessons, and finally talk about how I do the puzzle game, share some of my own strategies, I hope that peers can use it.
But before we get started, it's important to say in advance that our priority is fun and non-random puzzle games. You can also subvert some of the rules mentioned here to get other effects, and there's so much to cover.
So, let's first define a definition: what makes a puzzle game interesting? My definition is that when players play, as long as they show mastery, skill, or wit, we'll talk about these things later.
There are also gameplay mechanics, such as chests you can push and pull in the game, enemies and other similar things, most developers hear the gameplay mechanics, they will think of the scene on the left side of the picture, but we are talking about the right side today.
Part I: Human Running Memory
Human running memory refers to the ability to process information that you can memorize and use to solve puzzles. This metaphor may be more abstract, if your working memory can be calculated from 0 to 100%, what you do will occupy different memory, such as solving a puzzle may occupy a part of the memory, talking with friends, eating, etc., will occupy your memory, if there are too many things at the same time, you may not be able to cope.
Flow state refers to the fact that if the game is not demanding on skills but the challenge is too high, it will make the player feel anxious, if the requirements for skills are too high but there are no challenges, it will make the player feel bored, and between the two states is the flow state. While the specific proportions may not be very accurate, it's a metaphor, but at least we should take into account this aspect, such as how much memory the player needs to take up in the game I designed.
Conceptual compression
After many repetitions, the puzzle concept will occupy less memory space, such as rule A occupies 40% of the space, rule B, rule C each occupies a certain proportion, after a period of play, each of these rules will occupy less working memory. This gives players more leeway to learn new mechanics, and that's how the game works.
This is a novice puzzle solver in The Witness, which may require a lot of learning for players who have not been exposed to puzzle solving, but for a skilled player, they may be easy to pass, because each rule occupies less memory space and is just enough to show their puzzle solving skills, which is a very interesting balance point.
For an "expert" user, the purple and yellow rules may only take up such a little memory, but for novices, it may occupy more than half of the memory, so what is the extra? That's the space of possibility, and it shrinks as the player's experience grows.
In this puzzle, veteran players can see at a glance that a vertical line needs to be drawn in the middle, while for novices, they may try superfluous possibilities, which takes up more running memory.
Part II: Characteristics of Puzzle Solving
First of all, "Space-Time Illusion" is not equal to Sudoku, which may be a saying that many people have not heard, but why? Many people are probably curious, but we will speak slowly.
The difference between them is between hand-crafted puzzles and procedurally generated puzzles, and of course, there are many types of puzzles, but I think the easiest to distinguish is these two.
Hand-crafted puzzles are limited because they require players to repeat some tricks, and this type of puzzle can have more innovative solutions, and the puzzle content is officially produced by the developer. For players, this type of puzzle is more diverse and gives people a stronger sense of freshness.
Procedurally generated puzzles can have an infinite number of puzzles, players are repeating the same tricks over and over again, and you won't learn new skills, just like Sudoku, which is relatively uninvalued with no innovative solutions. From the content creation point of view, puzzles are generated by the system, I have been engaged in these two types of puzzle game development, for game planning, you only need to use a random generator. In addition, the experience that this type of puzzle will bring to the player is predictable.
Of course, in addition to these two types of puzzles, there are a lot of puzzle games in between, "The Witness" is one of the more interesting, its levels are handmade, but the puzzles in it are mostly generated programmatically, so it is closer to the procedural generation puzzle.
As mentioned earlier, our main focus is on hand-crafted puzzles, and linelight, which we released, is one such puzzle game.
Noise
If you can learn anything from my talk, it's probably the most important thing. My definition of noise is that anything that takes up running memory but doesn't contribute anything to solving puzzles steals (occupies) the player's running memory.
It's a puzzle from Linelight called Fliprs and Moving Streets. You're shuttling left and right in the middle of a hamburger-like object, and the balance beam on the right moves up and down.
This is another level of Linelight that looks a little more noisy, and its rule is to get three flippers at the right setting to complete the route, and that's the solution. It's just that it seems to have too many distractions, so if we remove these noises, the puzzle actually looks like this:
You only have to click on these two flippers to complete the route, so the rest of the design is noise.
This is a puzzle in The Witness, players need to find a way out from the lower left corner, in fact, all the paths here are noise, if you remove these noises, it looks like a puzzle on the right, it is easier to solve.
Therefore, because the picture on the left has too much noise, it occupies a lot of the player's running memory, and after the noise is removed on the right side, the memory occupied is much less.
The image below is a puzzle in Linelight that no one can accidentally solve because there is noise in this level. In this case, I deliberately designed these noises so that the player could not solve the puzzle by accidental attempts.
The puzzle rule here is to go to the top to unlock the path, can you see what the noise is? It's actually the loop below, and without these noises, people will just go up or down, and the puzzles will be easily solved by accident, and I don't know how they are solved.
Some people say that the fun of puzzle games should be to let the player find a solution unexpectedly without knowing it, but I don't agree, so I deliberately increase the noise to prevent this from happening.
This is an example of Inside, where the player needs to push the box over and let the rope on the left fall before continuing. If you push to the right, the rope will fall, and if you push to the left, it will rise. So you can think again, what is the noise in this level?
Yes, you can jump straight up to the box, which isn't part of the puzzle scheme for that level, but by definition, that's noise.
Overall, the appearance of these noises makes the puzzle more difficult, of course, the game planner needs to control the noise properly, if there is too much, it is easy for the player not to know what the puzzle is. And noise requires diligence to discern, and if noise isn't checked, it can affect the game experience you've designed.
Some developers may inadvertently design some noise, which does make the puzzle more difficult, but it is actually irrelevant to the puzzle itself, which makes it easy for the player to ignore the puzzle and only see these noises. If intentional, noise may help with the puzzle-solving experience, such as preventing the player from accidentally solving puzzles, which can lead to greater satisfaction if the player sees noise in the level design through skill.
7 lessons learned from Linelight
Article 1: Simplification
Less noise will make your puzzles cleaner and more compact. So how? Describe the solution to the puzzle and then remove everything you didn't mention. Reduces noise in the level design, making it easier for players to solve puzzles through practice.
In designing Linelight, I disassembled almost every level, removed all the irrelevant content, and then kept removing a lot of stuff, because we tended to habitually ignore some unnecessary elements when we made the game.
Now, the game can actually simplify a lot more, so my advice is to check each level at least twice to see if unnecessary noise remains in the level, and I actually measured each level more than 4 times during the LineLight design process. If some elements don't belong to a level, just cut them out, and it's satisfying for me to simplify the puzzle design.
If you want to increase noise in puzzle levels, remove the noise before intentionally increasing the effect you want. Because noise isn't exactly negative, sometimes it enhances the puzzle-solving experience, prevents accidental puzzle solving, and can sometimes be used to control or change the pace of the game.
This is the case in Linelight, and on the left is the original design, which actually looked very concise at the time, and I was thinking, what else can I remove? On the right is a level that has been stripped of a lot of things and redesigned, and the solution between the two is the same.
In another case, there is also too much noise on the left side, and the right side looks more visually pleasing.
In this case, I initially liked the puzzle solution on the left because it allowed me to discover a new way to solve the puzzle, but on the right side, the design was finally very simple, although the purpose of the puzzle was still the same, but from the left to the right version, I did many iterations.
Rule 2: Cut down meaningless levels
There must be such a level, and we have to accept this fact. So, for each level, ask: What is its purpose? If you can't answer this question, it means that the level is more mediocre.
There are also levels that may have a purpose, such as teaching the player something, enhancing a skill, or having unique moments that give the player an unprecedented experience, while others are just trying to mediate the atmosphere or break the rhythm of the previous game.
For the mediocre level, I would say that you are good enough, don't be afraid not to be so good, do things anyway, and every expert comes from mediocrity.
Of course, acknowledging that a level doesn't make sense isn't easy, especially when there's no better level to replace it. Diagnose every level and don't let any of them go.
Here's an example, this is a level I cut out because it's not educational, it's not fun, it's not fun, and it's not unique enough. It was actually processed by the two levels on the left, and I took some ideas from each of the two levels and made a third one, but it wasn't fun enough, so I ended up cutting the level.
Article 3: You are a professional player
I often hear players say that the "card time" of the level in the game makes it very difficult to pass, and even if they have 1.5 seconds to do it, they still can't pass the level. You may feel that this is enough time to pass, but for the player, knowing the solution but not being able to execute it feels very bad.
So, if you have to listen when the player complains, my strategy is to actually play through the level once, if you can't even pass yourself, then reduce the difficulty of the game, of course, this is only for puzzle games, action games may not be applicable.
For example, I spent hundreds of hours playing the game and was pretty sure what the difficulty the average player could handle. I changed the level to make it easier, but there are still players complaining, at this time, they are right, you have to make adjustments all the time, it is not them who are wrong, but you, and your game is for the player, not for yourself.
Rule 4: Separate the action from the game
This one applies to the solving of a single solution, which may not be so important if there is a puzzle with many solutions, but it is important if there is only one way to pass the level or very few solutions.
Both the action and the puzzle are great, but it's not so good to design the two together at the same time, for example, you don't have to serve tuna and ice cream on a plate at the same time, and you don't design the action and the puzzle at the same moment.
Because the action adds randomness and is easy to distract the player, solving puzzles requires a sense of rhythm and coherence. Players need an idea they don't yet know, and if the level requires action or puzzle-solving skills, you should tell the player that if only 10% of players complain, it's actually a serious problem.
So my advice is to make action levels optional. If you're going to do action levels, don't mix the two together.
Rule 5: Make the solution clear
What does this mean? Let's start with some notable phenomena: If a player says, is it right to do so? Have I solved the puzzle? Or try the impossible, which means the solution isn't clear enough and players aren't sure what they're doing.
If the solution is ambiguous, it will cause the player to lose trust in the game.
This is a level in Linelight, which is mainly a trial and error for players, so there is no fun. On the far left, players need to try the solution, followed by walking to the right, where my idea is to make the player stop and think about whether the previous solution is correct, and then go up to try a new solution.
But in the actual experience, the players were constantly trying and making mistakes, because the puzzle solution was not clear enough, so in the end I cut the level.
Like I said before, separate the different elements and remove the extra factors. If you're doing a platformer, be sure to make all the jumps obvious enough to let the player know where to jump or not.
I recently saw a similar bug in a platformer made by a friend. In this case, for example, if you're not a veteran player, you don't know if the table at the dotted line in the middle can jump. You want to let the player know clearly whether the failure is not understanding the rules, or the skills need to be further practiced, if the puzzle solution is not clear enough, the player does not know why it is not passed, and will feel frustrated.
Rule 6: Player Trust
In puzzle games, players default to feeling that agility is not required. For example, many games will instill a lot of rules into the player at the beginning, but in subsequent levels, as long as the rules are broken once, then all the levels may require reaction ability. I often find out that I am very reluctant to adjust because it is a very painful process.
Linelight players believe that a lot of puzzle solving requires responsiveness, because some levels do it. There are actually only 2 such levels in the game, leading players to believe that 15 levels require quick response. Of course, I know that each level requires or doesn't require a quick response, but players don't know that their understanding is real.
Article 7: Comprehensive design and fun
I had this problem while working on Linelight. We started by defining fun, which is to show the player's proficiency, skill, or wit, and the comprehensive and detailed design takes into account every possibility. But sometimes, the two are incompatible.
This question has plagued me for a year. In "Space-Time Illusion" and "The Witness", the developers have taken the comprehensive design to the extreme, and sometimes sacrificed the more classic fun for system reasons. After a year of struggle, I decided to only pursue player fun, so it was just my personal choice, I just wanted to say that this situation exists and if you encounter a conflict between the two, you have to have your own priorities.
In addition, planning things that are interesting, players do not necessarily think so.
For example, in one level, I designed each of each of the scenarios to be unique, and you can see from the naming, which I think is unique, interesting, and there's a lot of diversity.
To make a long story short, players are tired of the same layout, even though each level has a unique concept, but at the time I didn't obviously realize the boredom. Now it seems that this design is indeed boring, but at the time of design, I was very interested, I just tried to change a certain detail after what will be different, but never thought that the whole experience is unsatisfactory. So, in the end I reduced the 8 levels to 3.
Part IV: How Do I Make Puzzle Games?
The first thing to say is that this is my method, which may not be suitable for everyone, and besides, what I like to do is manually designed puzzle solving, so these methods may not be suitable for procedurally generated puzzle solving games.
Also, I don't know everything, and there are many more skills we need to learn, so this is not the ultimate guide to puzzle game design, but it is all a method that is very valuable to me.
People often ask, "How did you come up with the idea of designing this level?" "Actually, it wasn't what I remembered, it was the mechanism that did it. My strategy is to create some mechanics, and then use those mechanics together, and then there's the enigma.
So, how do we get these mechanisms? Mainly through experience, intuition and luck. Plus, I like to create some multi-faceted mechanics that can interact in a variety of ways. There's also the mechanism that excites you, the mechanism that you want to do.
When choosing a mechanism, get rid of the duds and don't force yourself to do everything, like some that aren't fun and some that I can't do. Try different mechanics and you'll always find interesting gameplay.
This mechanism looks very boring, just a mixture of red and orange, but what if we mix it with blue and yellow? You'll get more diverse results, that's what it feels like to explore the mechanism, and you'll find more surprises.
For these mechanics, my goal is to produce a lot of puzzles with a small number of mechanisms. There are types of games that can constantly try out new mechanics, but I want to try new mechanics after exploring enough possibilities, and, being pleasantly surprised by your mechanics, that's a good signal that you've designed something cool.
Solve the problem sentence
Linelight's best puzzles revolve around a moment of escalation, and my goal in making this game is to do the simplest level and use this concept as a solution.
For example, in this level, the solution sentence is "use the length to make the synchronization move clockwise", basically the simpler the sentence, the better, when the player makes the synchronized object move clockwise, they will have a surprise moment.
Elegance is inevitable when designing a problem-solving sentence, and as we said before, simplicity itself is elegant.
The diversification of the layout will happen naturally, in fact, each puzzle sentence is like a magic secret language, the player knows this secret language, they can find a way to solve the puzzle, if not, they have to try more, such as the following are some of the cases I have used:
In fact, there are "shortcuts" to solving the problem, such as:
As you can see, it's rhythmic, but each level is still a completely unique challenge, they're scattered throughout the game, and sometimes small changes in the expression of the sentences often mean big changes in the layout. This is also what surprises me, in my opinion, the result of this is likely to be that the mechanisms on the left are inherently diverse.
However, people will still ask, how exactly did you get these sentences?
My approach is to force these mechanisms together and see what happens, and sometimes, when you see something happening, you can get the idea of solving the problem. Ask yourself "What if... What will happen? "Don't invest too much time trying, though, and let the system find the answer for you."
Doing puzzle solving should not be painful in itself, because good puzzle solving often comes naturally under the impetus of mechanisms.
Of course, there is also a more specific method, that is, to list different puzzle elements in a tabular manner and explore the solution sentences in a systematic way, which tends to be more efficient.
For manual puzzle games, no one's head is actually big enough to actually "come up" with these puzzles created by the planned system, and I am very sure that most puzzle games, the game planning creates the mechanics, rather than creating all these puzzles one by one.
In other words, the planner does not "give birth" to these puzzles, but helps them to arise, just like the midwife doctor in this picture, the left side is your game, and the puzzle is the product of the game mechanism.
Do puzzle solving with running memory
It takes 75% of the memory for a player to solve a puzzle, but for me, I thought that creating the puzzle might require more memory, but in fact, if there is a good mechanism, it is easier to build these puzzles than expected.