When the Pattern Leaps Out

The trick to Portal 2 community chambers is to make needless complication look clever and obvious in retrospect.

The computer game Portal 2 is a three-dimensional puzzle game that relies heavily on a well-established physics. The player's character has a portal gun, which allows the player to create two linked wormholes on solid surfaces and pass from one to the other. As you pass through, you maintain your momentum, but gravity acts differently on your body, so you can do a variety of interesting things with these facts alone: you can launch yourself great distances by leaping from a raised platform onto a lower horizontally-placed portal which is linked to a portal on a vertical surface. Gravity increases your momentum on this side of the portal; on the side of the portal gravity now starts pulling you in a direction orthogonal to your trajectory, but momentum carries you pretty far. However, there are a number of other devices in the puzzle chambers with you: lasers and laser catchers (which can trigger other devices), bridges made of solid light, lethal turrets with friendly and forgiving artificial intelligences, platforms on moving pistons, and so on. (Sixty Symbols has some videos about the Portal physics.) There is a single-player narrative arc (which is excellent, but I won't get into it here) and a cooperative narrative arc (which I haven't played), but what I want to discuss is are the community test chambers: players can design their own stand-alone chambers and allow others to play them.

Screenshot of part of the solution to one of the puzzles in the Portal 2 game.

Although the portals, gravity, momentum, and whatnot are all functionally analogue, the other gadgets are often binary. They are either on or they are off. Triggers are either triggered or not. A laser stream is either obstructed or it isn't. And most devices can be triggered by another device: light bridges can be switched off, swinging panels can be opened or closed, moving platforms can be activated or deactivated. These consequences are also all binary. As a consequence, you can make elaborate systems with these devices.

You could, for example, make a computer.

You couldn't make a terribly fancy computer. You have a limit to the number of devices you can place in a chamber. But user TtHsa-1 has made a computer. First, some background. TtHsa-1 has always made interesting systems: for instance, in Test Cycle - Concept the player can place a weighted cube on a button, which triggers a series of reactions that ultimately remove the cube from the button and then puts a new weighted cube on the button, starting the system over again.

That chamber is the one that got my attention because I was working on a similar chamber at the time: having seen this video about iodine clocks, I wanted to make a Portal 2 chamber that repeated itself but with some remainder that eventually absorbed the entire process. In other words, I was planning on making a cycling process which degraded with each cycle. But I could not figure it out; what I did figure out in the meantime was how to make a puzzle which ran in part on an automated system. Unlike TtHsa-1's, mine started automating as soon as you entered the room. (It involves a system I just called "the engine," which involves buttons, weighted cubes, and tractor beams.) Once the cycle begins, the chamber has two states which alternate automatically, and the player needs to learn what each states looks like and incorporate those states, and their alternation, in her solutions to the puzzle.

So I was surprised to see TtHsa-1 working on a similar project, though he makes systems, as I said, and not puzzles. But then he started working in another direction: logic gates. With the devices Portal 2 provides, logic gates are actually fairly easy to make, though some of them are device-expensive. And TtHsa-1 must be far cleverer than I am, at least about this sort of thing, because he made then made a 6-bit adding machine. It's pretty slick: you get twelve cubes and two rows of six buttons. Each row represents a line of binary, so you can make two numbers by placing the cubes on the buttons: a pressed button is a 1 and an unpressed button is a 0. The output display is on the wall, made of seven flipping panels: the black (unportable) side of each panel is a 0 and the white (portable) side of each panel is 1. Yet again, it reads as a line of binary. Between the buttons and the panels is a very complicated system of lasers and laser catchers which make the necessary logic gates.

TtHsa-1's chambers are very clever and they're also quite clean and beautiful. And I appreciate all of those qualities in a puzzle, so much so that most of my pleasure comes from those qualities. But there's something his systems don't have that puzzles do: the retroactive delight of finding out that a complicated system makes sense after all. A good puzzle is much like a detective story: what appeared to be meaningless unconnected pieces are in fact coherent, but only once that coherence is discovered. It's the moment the pattern leaps out at me, in its cleverness and efficiency and beauty, that I really love. The fact that I worked for the pattern helps on a psychological level, I'm sure. Some of these puzzles are very relaxing, because the pieces just sort of fall in place as I interact with them; others are much harder and require quite a lot of mental modelling in order to work out what I'm going to do before I start doing it. I enjoy both, but with the first kind I enjoy the creator's cleverness, while with the second kind I get to enjoy the creator's cleverness while also feeling clever myself. TtHsa-1's systems can't offer any element of the latter, though it offers quite a lot of the former pleasure.

I've been trying to think of ways to use his inventions in a puzzle. The most obvious is to create a 6-bit calculator and then hook up the exit to a particular answer, which you can show (in Arabic numbers) on the ceiling. But I've tinkering with a set of rooms which rely on manipulating the logic gates, or actually the same logic gate over and over again; the final room would require you to have figured out how the logic gate works, because you would have to build one using movable pieces. It's...not going so well.

Two parting, not entirely related, thoughts:

1. The devices are nice and fun, but some of the best puzzles hinge on something obvious but easy to forget. For instance, a lot of nice puzzles mostly use devices throughout, but the final stage requires you to remember that gravity exists, or that nothing can pass through solid objects. In fact, I find that when I design puzzles I often forget that gravity is an element of the puzzle which I can incorporate into my design. Sightlines are also an important part of puzzle design.

2. I really like puzzle designs in which the same chamber involves multiple puzzles, where the different puzzles use the same pieces. For instance, one of mine has four chambers, in succession; you must solve each on its own to get to the next one, but when you solve the final rooms you then have access to all of the rooms and must solve a final puzzle which involves elements of all four rooms. Since I like four-room chambers, I'm also thinking of making a kishotenketsu-based puzzle which repeats the same kind of puzzle in the first two rooms, uses a whole different design philosophy/puzzle type in the third room, and blends the two in the fourth. (I hear this is what the makers of the Mario games did.)