Designing the mind-bending puzzles in Patrick's Parabox

Patrick Traynor’s latest puzzle game Patrick’s Parabox is both an evolution of the block-pushing puzzle game, and something new and unique to itself. It allows players to shrink down and enter boxes, where there can be whole levels inside them, but that is just barely scraping the surface of a unique and ingenious experience.

According to the game’s promotional text, every puzzle serves to communicate or reinforce some new concept, there is no filler! In a way reminiscent of Baba Is You, things keep on getting weirder, and the player is left wanting to keep going just to see what craziness will happen next. Patrick’s Parabox won the won the 2019 IndieCade Developers Choice Award, and Excellence in Design award from the Independent Games Festival in 2020.

Joel Couture did an interview with Patrick a couple of years ago while Parabox was in development. We caught up with him about a month after the game’s release and asked him some more questions, which he graciously answered.

Who are you, and what is Patrick's Parabox?

Hi, I’m Patrick, I’m an independent game developer and programmer from southern California. I’ve made a lot of freeware games, including many precision platformer fangames of I Wanna Be The Guy including I Wanna Maker, puzzle games, game jam games, and a few game dev tools. In recent years I’ve been super absorbed with playing and making puzzle games!

Patrick’s Parabox is my first commercial game. (The name is inspired by the game Stephen’s Sausage Roll.) It’s a block pushing puzzle game with a recursive twist: boxes can contain their own little areas, which in turn can contain other boxes, which can contain their own little areas with more boxes, etc. It’s quite a mind-bending system, and it gives rise to lots of mechanics and puzzles, and so it turned into a commercial game.

So, to introduce the game to readers, feel free to correct me if I misspeak—The simplest explanation is, Patrick's Parabox is a Sokoban-like puzzle game where the boxes you push, themselves, can contain areas, with their own boxes and goal squares inside them, and you can enter the boxes in some circumstances, where you shrink down and the screen zooms in, then the world expands when you emerge. Those are the briefest basics, but the game introduces a stream of weird new ideas and rules, and to flirt with understatement, things get a whole lot crazier. Without spoiling things too much, is there a favorite example you could give of how crazy they get?

Yeah, the game kind of takes the premise and runs wild with it, exploring many consequences, edge cases, and twists. One fun example is, there are scenarios where a box recursively contains itself—this takes a bit of practice to wrap your head around! But then, what if you took that box (that contains itself), and pushed it inside of a different box? The structure of the game world then changes in a kind of abstract and interesting way, that’s hard to imagine until you see it and play with it. And yes, there are many more recursive wrinkles introduced later on, that are better left as surprises!

You could think of Patrick's Parabox as being about Sokoban, a series of riffs on it. Some levels take the form of offering something a situation from out of Sokoban, like a needed block being stuck in a place you can't get it out from, but then asking the player to get the block out using the level's gimmick to subvert that. The influence of Hiroyuki Imabayashi's Sokoban is so wide-ranging. A brief and by-no-means complete of games with Sokoban elements: Eggerland/Adventures of Lolo, The Legend of Zelda, Chip's Challenge, Stephen's Sausage Roll, Snakebird, Baba Is You, and, strangely, freaking NetHack. It's basically the default computer puzzle game. Why do you suppose it's such a fertile ground for variation?

Great question! This is fascinating to me too. In my head, I summarize it as: discretizing positions and state to the square grid, and interacting with objects in a physical manner on that grid, ends up being a versatile base for puzzle systems with rich interactions, while at the same time being intuitive for us humans to reason about. I’m not knowledgeable about why this is at a low level, but I would love to read some kind of analysis of it!

To muse a little more: The physicality of the interactions and objects (rolling a log, spearing a sausage, reshaping a snakebird) surely has something to do with it. There’s so many real-life phenomena to draw from and adapt to a sokoban-like. And the resulting mechanics can feel grounded in reality, which might enhance that magical feeling when something extraordinary happens, like pushing a raft, or shrinking down to enter a box. Also, I feel the basic Sokoban premise of pushing crates around solid walls has some kind of inherent mathematical or computational complexity that makes it a viable core to expand upon.

Patrick's Parabox feels of a piece with Hempuli's Baba Is You, in that the game is constantly introducing bizarre new twists on the premise, and challenging what you thought you know about it. How do you feel about that comparison? Is it accurate?

Yeah, definitely an accurate comparison in that sense. I think both Baba and Parabox are based around core systems with similar flavors—abstract, modular, powerful, and with a certain special mind-bending feel to wrapping your head around it, like learning to “think with portals” in Portal. Both games are also fairly diligent about exploring the depth of their respective systems, making puzzles out of edge cases, and adding natural extensions of the mechanics. It’s delightful when puzzle games have these twists, and it was a joy discovering them myself when when working on Parabox! I attribute these twists to how inherently deep or versatile these puzzle systems are.

Patrick's Parabox's elegance masks a game that took you years to make. Something I wonder about games like this and Baba Is You is, how did you come up with the new rules? Like, when shrinking down to enter a box, there must be room in the middle of the side you're entering. Things like that, and allowing the player to push rows of boxes at once. While playing, I get the sense that these nuances were chosen with care. How do you iterate through the implications of rules like that?

The rules were certainly chosen through a lot of care and iteration! For example, in an early implementation, if there was no room at the middle of the side of a box you want to enter, the game would then check for room at all of the other tiles of that edge, checking spots closer to the middle first. I then realized this was pretty inelegant, and would make for awkward puzzles utilizing this behavior, and would be an awkward rule to learn and remember. The current center-only behavior felt a lot better.

I put a lot of thought and iteration into how to resolve many other scenarios too! Here’s some principles I followed when making rule decisions:

1. Prioritize natural/organic behavior of the system. Avoid arbitrary choices or artificial simplifications. The talk “Truth in Game Design” by Jon Blow influenced me a lot.

2. Prefer the behavior that makes more interesting puzzles.

3. Prefer intuitive behaviors, and behaviors which are easier to remember and reason about.

4. Implementation difficulty was a factor in some decisions; some things were [not feasible] for me to implement.

I wonder if you must muse about the puzzles not made, the gimmicks made moot by doing things in a particular way. Did you get to a point where you wished you could have gone back and made a different choice?

I’m pretty happy with most of the behavior choices I made. There [are] definitely a few awkward spots, but they’re pretty rare to encounter, and I ended up avoiding them for the most part in the level design. I’ve mostly made my peace with them, haha.

Minor spoilers for the game in this paragraph: I do have to mention, there is a small part of the game which touches on this exact idea, and explores what would happen if a few rules were chosen differently! I think it turned out pretty well, and it definitely scratched my exhaustive design itch. I’m interested in exploring this “alternate rulesets” idea in future games, and I’d love to see it in other puzzle games too.

Puzzle games of the same kind as Patrick's Parabox aren't just puzzles, but they also teach you how to play, by stages and without words, as you pass through them. Something about this style of game that worries me, when playing, is the possibility that I might stumble upon a solution without understanding what I did, and thus not be prepared for later puzzles. Or I might come back to the game after a break, forget the thread of previous levels, and immediately hit a wall. Is that a concern for you too? What is it about this teaching-by-doing that makes puzzle games of this sort compelling?

I made the puzzles in Parabox follow my personal taste and playstyle—I enjoy easy, flow-y puzzles that still have interesting ideas, and I sometimes bounce off of difficult puzzles. So in general my philosophy was “make the easiest possible version of a puzzle, that still features the interesting idea or point, and still showcases it or teaches it with a lot of impact.”

It was a balance. In frequent cases, I decided I was okay with people totally stumbling upon a solution and not understanding it at all, so long as they properly grappled with that same concept in the immediately following puzzles. I was also okay with people stumbling upon the solution by accident, but retroactively understanding why it worked, for concepts that aren’t vital later in the game. It’s important to note this philosophy definitely isn’t watertight—people do end up underprepared at points, and sometimes don’t fully understand certain concepts even after completing the game. Even so, I think [it] flows well for a lot of people, and it’s great to see the game’s friendly difficulty appeal to non-puzzle-expert players.

One thing I admire about Patrick's Parabox is the utility of the graphics. For example, when you enter a box, the camera zooms in, but still shows the context around the box, aiding the player's memory and also immediately explaining why they might not be able to leave in a direction, as a blocking wall is still visible. Also, there are the animations illustrating what happens when entering/exiting a box, which seem like they must take into account a lot of weird situations like going directly from inside one box to another, or illustrating infinity. You mentioned in a previous interview that the graphics were very difficult to get right, and at first you thought they were kind of "programmer's graphics," but they came to appeal to you because of their abstract nature fitting the game. I agree that they're very nice for the kind of game it is. Did you finally manage to sort out all those edge cases? Are there any regrets concerning its appearance?

Thanks for appreciating all those little touches! Many numbers were tweaked, much thought was put into handling and visually communicating edge cases, and many visual effects were experimented with. The box motion was perhaps the most challenging section of the game to program—like you said, there are edge cases abound involving recursive situations. There are a few tricky ones that were left unsolved, but in the end they ended up being good enough or even unnoticeable. I think that’s typical fare for game dev, haha. As for overall visual style, I’m sure there better possible iterations, but I’m okay with how it ended up.

Do you ever wish you could have included even more puzzles, or do you maybe think you put in too many? How do you decide how much content is enough?

I followed that old philosophy that a game should be as big as it “wants” to be. Which is quite vague, but basically means aiming for what feels like a natural length based on the premise of the game—not under-exploring, but also not overstaying its welcome. It’s hard for me to judge Parabox since I’ve been so close to it for so long, but I’m pretty happy with its length. It’s possible the game is a little verbose, but on the flip side there may also be some untapped potential for making more complex puzzles.

I was pretty ruthless when it came to iterating on puzzles. At times, big batches of puzzles were added; but at other times, big batches were removed. (And many times, puzzles were modified!) There’s this quote I like: “Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away.” The game is of course far from perfect, and I also started to see diminishing returns on this iteration after a time, but I think it ended up in a good spot.

An amazing thing for a puzzle game like this, but something that feels entirely appropriate due to the nature of the game, is the check box in the options screen that says "Unlock all puzzles." Just, right there, offering to give the player all the puzzles in the game without rigamarole, and also allowing them to turn it off. I have viewed video and computer games' willingness to "lock" features away from people with limited time, who bought the game after all, with suspicion. Yet, of course, a player who tries a late-game puzzle without going through the puzzle before it will be entirely lost. What caused you to offer this feature, and what does it mean, do you think, for the way the player approaches the game?

I took this feature from the game Cosmic Express! I like that game, but personally I sometimes bounce off of difficult puzzles, and I got a bit stuck at points. I enjoyed being able to see the whole world map, and check out mechanics and puzzles, even if I wouldn’t actually solve most of them myself. In Parabox, I think it’s less useful because the game is lenient in letting you progress, but it’s still a good feature in principle. Maybe you lost your save file, but you still want to pick up where you left off, or you want to show a friend a certain area of the game. It was also a useful feature for my own testing during development!

Are you tired of people referring to the movie Inception when talking about Patrick's Parabox?

Yes, but I’ve gotten used to it, haha.