Main blog page


Are mazes bad game design? Why do so many players seem to think so? Lately I've been pondering these questions. I'm not a game or level designer, but I have worked in games for a while as a programmer. Which means I get to have opinions, I guess.

Anyway: over a decade ago, I read an article from Notch giving his take on this: "the player has no way of knowing or figuring out before hand what decision is the correct decision. Don’t penalize the player for things they can’t control".

This was before Notch got a billion dollars and rotted his brain

That seems like a decent explanation, and was probably the first time I thought about the matter. It stuck in my head as "maze bad". But more recently, I started to wonder...what even is a maze? I think there are some "maze-like qualities" that we can identify:

Non-linear structure: A wholly linear level cannot be a maze. There need to be branching paths at some point. There is some interesting history here: labyrinth traditionally referred to unicursal, or single-path mazes, but in modern English we usually labyrinth and maze to be equivalent.

Branches must have dead ends: i.e. they must require backtracking. A common level design trick is to have branching paths that then reconvene or merge together. This is discussed more in this excellent video by designer Peter Field, who calls it "the illusion of choice". It's part of a common philosophy in games which is that it's often cheaper to give players an impression of something rather than actually giving them that thing.

That does introduce an interesting follow-up question: how many dead ends do you need for it to be a maze? There's not a hard line, but I think if you kept adding them, you'd reach a critical point where most people go from saying "it's a linear level with side paths" to "it's a maze"

Branches cannot reveal the correct path until after the player has chosen: Plenty of dungeons have little side paths or alcoves, things that are obviously not the main path. It's not enough to have non-linear structure, you must have decisions where the choice is arbitrary, and no matter how hard you stare at it, you won't know if your choice is right until after you make it. This is the criticism voiced by Notch earlier, that you're arbitrarily punishing the player.

As an alternative, think about a multiple choice quiz minigame: a series of questions. You could just guess and check your answers, like in a maze. But the point is that you're supposed to have some idea of the right choices, based on your existing knowledge

No mini-map: This builds off the previous one. There's some discussion on a 2003 forum where players talk about how mazes are getting to be less common (even back then). But one thing they say is that newer games have an "automap" instead of requiring manually charting them. A maze with an overhead view is much easier to solve than a 1st person view. So much so that people feel they're no longer mazes, just regular non-linear levels.

Destination Unknown: Not always, but plenty of mazes don't even show you where you should be going long-term. You're meant to wander around aimlessly, breadth-first searching without even the benefit of A*.

Little visual variety: Think about the city you live in: it's non-linear, branching, but it (probably) doesn't feel like a maze even without GPS. That's because every place feels different, you can build up a mental map. Usually, knowing one part of a city will also help you navigate other neighborhoods, your brain picks up on patterns and rules for wayfinding. With most mazes, each position looks similar, if not the same. Homogenous walls, little visible structure. Also, cities usually have much longer sightlines, making it easier to see the bigger picture. Contrast this with mazes, which are often downright claustrophobic.

I haven't read it, but Kevin Lynch's "The Image of the City" is supposed to go much more into how we navigate cities

Raph Koster posited a theory of what makes a game "fun": that fun is synonymous with learning. And eventually, when you master a game, it gets boring. You'd think in this way, mazes provide infinite fun: no matter how many you do, you never get better at them. Not really anyways. It's always still a 50-50 guess at each fork: left or right? You're constantly learning in the sense of "when I'm at this specific location, I should turn left", but these are all only pertinent in that one context. There's no generalization.

I don't know enough about psychology, but I suspect our brains pick up on this. They can tell that it's pointless, that we are brute-force, rote memorising rather than learning to see patterns. Even worse, mazes are artificial, they lack all the affordances of real-world spaces. We don't feel like we're exploring a space, we feel like we're solving a maze.

But still, people do dumb pointless things all the time, right? In middle school, I memorised 250 digits of pi, as a substitute for having a personality. And it's not like knowing the first 249 digits of pi gave me any clues about the 250th one. There's no pattern, no generalization, just like mazes.

inb4 some mathematician comments that pi is not known to be a normal number, so maybe there is some pattern in there

Further, if mazes cause frustration, confusion, and disempowerment, doesn't that fit right in with some games? Survival horror like Silent Hill, but even just games wanting to make a punishing atmosphere: Fear and Hunger deliberately hides a lot of information from the player. Famously, there's a toilet which, if interacted with, asks if you want to jump into it. Going through with it soft locks you. This is just like a maze, but even more punishing: instead of backtracking, you have to reset your run.

And yet, there are still rules. There is still an underlying logic to the world, even if it can feel arbitrary sometimes. Fear and Hunger has to balance rewarding exploration and creativity with its oppressive vibes. But players do make progress, they do learn and get better at the game. Similarly, with Silent Hill, you don't just go from point A to point B, you actually move through it like a real city, accomplishing goals along the way. The game gives you reasons to navigate the world, and at least some of the affordances to make it feel natural (if also confusing and creepy).

But even after all that, I'm not ready to give up on the maze entirely. I don't think they should be the first, second, or third tool a designer reaches for. But, maybe they're worth keeping in mind. Maybe there's a place for things all along the "maze spectrum" which I'm coining to mean everything from corridor shooters to full-on Prim's Algorithm. That talk by Peter Field I linked earlier mentioned that it's important to sometimes let the player get a bit lost. Like so many things in art, it's about building up tension and releasing it.

Games that move down the maze spectrum sacrifice the dopamine hit of meaningful learning in order to unsettle the player. A lot of old adventure games (where I suspect their reputation comes from) didn't make this decision intentionally, and it was more about padding playtime. Especially if you don't have other mechanics that drive player progression, mazes feel clunky. If you're going to deny players the satisfaction of mastery, know what you're doing and why. Or at least throw in a "Skip Maze" button like Memoria did.

"To create is to suffer. To design is to find some meaning in the suffering." -The demonic spirits in my mind