What Makes a Good Puzzle?

Puzzles can be one of the most complex things in game design. In this video, I break down some great conundrums from favourite games, and share some knowledge from sharp puzzle designers, to find out what makes a good puzzle.

=== Sources and Resources ===

Level Design Workshop: Solving Puzzle Design | GDC

A Good Puzzle Game Is Hard To Build | Rock, Paper, Shotun

=== Chapters ===

00:00 - Intro
01:00 - The Mechanics
02:37 - The Catch
04:28 - The Revelation
08:01 - The Assumption
11:03 - The Presentation
14:49 - The Curve
15:55 - Conclusion
17:07 - Patreon Credits

=== Games Shown ===

Snakebird (2015)
Braid (2008)
Portal (2007)
Stephen's Sausage Roll (2016)
Induction (2016)
Yono and the Celestial Elephants (2017)
The Talos Principle (2014)
Cosmic Express (2017)
Deus Ex GO (2016)
Pipe Push Paradise (2018)
Inside (2016)
Lara Croft GO (2015)
The Swapper (2013)
Portal 2 (2011)
The Misadventures of P.B. Winterbottom (2010)
The Legend of Zelda: Skyward Sword (2011)
Hue (2016)
The Turing Test (2016)
Black the Fall (2017)
Rise of the Tomb Raider (2015)
Uncharted 4: A Thief's End (2016)
Ittle Dew 2 (2016)
Agatha Christie - The ABC Murders (2016)

=== Credits ===

Music used in this episode:

Snakebird OST
k. Part 2 - 01 untitled 1, animeistrash

=== Subtitles ===

[DoubleATam]

This was mentioned in passing on the video, but just to re-emphasize: not every puzzle should be that hard. You might want a couple easier puzzles between hard ones, to not only make the player feel smart, but to build assumptions to later subvert.

[KoolKojoS]

While I was watching this video, I literally paused it when he got to SnakeBird level 10, went on the app store to see if it was even out on iOS. Found it, played up to level 10, got stuck for about 20 minutes, figured it out, and continued the video only to see Mark explain the level in the EXACT sequence that I went through unconsciously. Great video Mark! :)

[xisumavoid]

The snakebird soundtrack is fantastic :-)

[stevenvass]

Mark, one thing you didn't cover is giving the player the choice of several puzzles at once. This way if they get stuck on one, there are others to attempt. Snakebird's map is a great example of this, as is Infinifactory, The Witness, and Stephen's Sausage Roll. Portal 2 used a linear design which greatly restricted player choices, but they spent thousands of hours playtesting to make sure it would work. For an indie puzzle dev, I think giving your player's multiple puzzles to attempt at the same time (outside of simple puzzles that introduce new mechanics) is a winning design. Absolutely loved this video, and wish this one tidbit was mentioned because it would make this video "complete", in my opinion.

[NintendoCapriSun]

I remember thinking I could design puzzles when I was a kid. Found out pretty quickly that I was wrong!

[LinkingYellow]

This was great! If I could add anything, it'd be how important it is that the correct solution works the first time it is tried. I've created too many puzzles where the player tried the correct solution, messed up with execution, and wrote it off as the incorrect solution only to get stuck till they give up.

[mothersbasement]

Fantastic video. Puzzle design is fascinating and I'm hapoy to see you finally tackle it. One thing I will add is that if you're not careful, you can create a predictable rhythm of assumptions and "catches" that players catch onto over the course of your game. I remember playing through the room (fantastic game btw) and reaching a safe puzzle where I had to turn a dial in the correct sequence of directions to open a door. It's meant to be a moment of trial and error where you find the sequence by getting it wrong a few times and correcting your mistakes, but I was able to figure the sequence on my first try just by second guessing the designers based on patterns in previous puzzles.

So I guess the final, broader tip I'd add to this video is to think of your whole game as a puzzle in itself, and remember that players are building up assumptions across their entire run that you need to subvert.

[ShinoSarna]

Interesting idea I've been thinking about : there are actually TWO types of puzzles.
There are 'ruleset puzzles' and 'adventure puzzles'.
Ruleset puzzles are the ones you discuss in the video - they exist mostly in puzzle games - in these, player knows the rules of the game, and the trick is understanding what sequence of moves will bring them to the ending (perhaps while pushing logic of the ruleset to its absolute logical conclusion). The trick to those is being able to visualize the puzzle in your head, and thinking several moves ahead, and thinking deeply about what you already know about the ruleset.

In adventure puzzles (as name suggests, mostly found in adventure games) where player DOESN'T know the rules of the puzzles, but puzzle is absolutely trivial once you do, and figuring out the rules is the entire point - e.g in Sam n Max S1E1, player is tasked with finding some swiss cheese for a rat, and has availiable a gun, and a giant pile of plain gouda cheese. Once player realizes the rules of the puzzle (that bullet holes in cheese can make it look like swiss cheese), solution becomes trivial, and can usually be solved in one-two moves. The trick to these is lateral thinking, to try to apply real-life logic to try to find out what the rules are.

Of course, there are also best practices for designing adventure puzzles - for example, making sure that the solution does follow either real life logic or in-world logic (otherwise we get 'moon logic puzzles'), that player KNOWS about all objects needed to solve the puzzle, to telegraph the solution in case of more obscure puzzles (basically, add some sort of clues or hints), and also making sure that there isn't e.g. a more logical way to solve the puzzle at hand that the designer didn't account for (if player has a retractable ladder in their inventory, asking them to climb a ledge by creating a grappling hook from a piece of yarn and a fishing hook might leave players seriously confused); and finally, NOT asking players to use outside world knowledge. If player needs to look some fact up to progress the puzzle, let them look it up INSIDE the game (Blackwell games did this beatifully with its in-game search engine).

The interesting part is that some people did experiment with trying to use BOTH types in their games - for example, in Witness, Johnattan Blow regularly changes rules of the puzzle, and figuring out the rules is very often an objective of the very first puzzle in the set - and sometimes it just goes for a straight up 'adventure' lateral thinking puzzle (like with the 'apple in the background' puzzle - if you played Witness, you likely know which one I mean.)
Of course, adventure games also often include occasional 'ruleset' puzzles to add some variety (often in some form of pipe/ventilation puzzle, or a puzzlebox holding a key item), but that's usually confined to one puzzle in the game, and can feel arbitrary and annoying if done incorrectly.

[ZiggyStyley]

I remember watching 'Teens React to Portal' and finding myself being frustrated watching people struggle with the first puzzles, but this gave me an appreciation for how the game teaches you to change your understanding of how mechanics work. The kids struggled because they were new to the game and were looking at the puzzle with a normative logic, once they began to get their heads around the mechanics of portal solutions became logical. It became a real pleasure to watch people learn to understand the game for the first time.

[carterbrown]

Another thing to remember when making a puzzle game is that you have to want to continue. If it is just level after level of puzzles it can easily get boring. In the first half of Portal 2, you want to keep playing to see Wheatley again. The setting, music and entire mood of the game makes you feel lonely, which makes you want to continue just to find someone to talk to.

[Dilandau]

As Myst designer Rand Miller said, a good puzzle is one where when you look up the answer you go "of course, why didn't I think of that!?" If instead you go "I would never have thought of that" or "that's a stupid thing to do" the puzzle designer has failed. Of course, this helps you evaluate whether a puzzle is good, not design one. Games like Myst make things even harder for the designers because they're trying to present a believable world as well as a puzzle game. The elements of the puzzle must feel like it makes sense for them to exist in the world besides just serving as a puzzle for the player, which adds a layer of complexity. I feel the best puzzles are those where you don't feel like you're solving puzzles, but you're just trying to figure out how the world, or some piece of it (e.g. some contraption) works and that happens to be a puzzle.

My biggest pet peeve is puzzles that are easy to solve, but where the solution is difficult to execute. It means that you can't easily attempt any ideas you have on how to solve it. "I think this is the solution, but finding out if I'm right will take me ten minutes" is never a good thought to have during a game.

[SDGGames]

6:25 I remember playing a Zelda game for the DS when I was a kid. At one point, you needed to copy something from a map on the top screen to your personal map on the touch screen. I poked at it for an hour with no idea what to do, but when I came back from dinner later, the puzzle had apparently solved itself. When I played through the game a second time, I got stuck at the same exact place. At one point, I closed the lid of the DS, and finally realized that touching the two screens together was actually the solution. I still remember the frustration I felt in that moment to this day. If solving a puzzle doesn't cause any euphoria, it's probably too contrived...

[Gilboron]

I actually found this pretty interesting to watch as a writer, too. After all, many challenges or mysteries in stories are essentially a kind of puzzle. Of course, there's differing factors at play there, as it's the characters who need to solve the "puzzle" and the finesse lies mostly in making the reader/viewer feel involved in the solving of the "puzzle" without making the characters seem slow or stupid. But there's still interesting correlations to make between a good video game puzzle, and a good narrative puzzle.

[jebussonofgob]

I'm so happy to see Snake Bird getting some love. Such a good puzzle game.

[ruolbu]

But HOW do you get those three sausages grilled and reach the exit?!?

[romainbajoit]

"Every puzzle starts with its mechanics : a set of ironclad rules that governs how the game works .."

*BABA IS YOU:* HOLD MY BEER

[OutbackCatgirl]

There's an aspect to puzzle games that I am deeply in love with that may not have been touched on here... Unintended Solutions and *intended* 'unintended' solutions. Where there's an alternate method of completing the puzzle that feels like you're "breaking the rules" to attain it, often involving more time and skill, being technically challenging to achieve, often giving - in my experience - a happy little thrill, when most of the time that solution will have been left in *intentionally* for one reason or another, like that one chamber in Portal where you can skip directly to the top of the exit elevator with a single fling, making the player feel extra smart for 'outwitting' the puzzle's boundaries.

Or most of the gold stars or easter eggs in Talos Principle which involve smuggling puzzle elements outside their designated areas. (Talos is such a phenomenal game imo, it gave me that very feeling so many times, and is the first puzzle game in years that's prompted me to not only go for a 100% completion run, but to also get all endings and bonus objectives and even easter eggs.)

[StefanLopuszanski]

As a puzzle designer myself, I think one of the best things I've learned is you want to create a puzzle that is like a good math problem. That seems weird, but hear me out.

Puzzle games are like teaching people in the education system. You explain to them the basics, go over all the basic rules and goals, and then give them problems to solve using the things you've taught them. Puzzle games almost universally have a set of rules and interlocking mechanisms that are used in a specific series to overcome a challenge.

This is also how math problems are solved. You know the rules and tools you have and need to figure out which to use and what order to use them in to get the solution.

If you don't properly teach the rules and mechanisms then players will not be adequately equipped to solve them. Just like if an educator didn't properly prepare students then the students will have issues. Also, you can't give a student (player) something too complex at the start or they will get frustrated and give up. Having a good onboarding process is crucial for puzzle games. You might have the best puzzle game mechanisms out there, but if players can't grok it then it isn't going to do well.

[BRICK]

Great video! That Talos Principle puzzle is a great example. I remember that moment really clearly, it was one of the most well designed moments of the game

[SavageGreywolf]

In a great puzzle, you immediately see the goal, spend the majority of your time in the level figuring out what you need to do to get to the goal, and once you figure it out you quickly are able to achieve the goal condition.

In a bad puzzle, you spend the majority of your time in the level trying to find the goal, once you find it you quickly are able to figure out what you need to get to the goal, and immediately afterward achieve the goal condition.

In a TERRIBLE puzzle, you quickly see the goal, you immediately realize what you need to do to get to the goal, and you spend the majority of your time in the level trying to achieve the goal condition.