Words by Hamish Todd.
Portal has the best-designed first-person puzzles I've ever seen. They’re surprising, focused, and concise. They are also designed very perceptively, and we can learn a lot from looking at this perceptiveness. Read on for an analysis of Portal's level design, and some lessons about what learning from it can do to improve game design.
BE WARNED: This article uses multiple animated .gif images on the same page, and might be tough to load on slower connections.
This diagram shows a move called the “fling”, where you turn a fall into a leap forward using portals. It is a beautiful and interesting possibility within the game engine, and the level design is intelligently constructed to make it accessible. On the surface, Portal is a game about changing your position in space, but the fling move is about changing your direction and repurposing gravity as a means of increasing your elevation. So the fling move is a quite sophisticated idea; it’s also rich in uses, and involves you moving in a swift, continuous curve - so it’s very fun.
A fling is also an extremely disorienting experience. This makes it cool to plan around, but difficult to teach. Good thing then that Valve’s level designers are expert teachers, who know exactly how to craft around player’s thought processes. The sequences in the game that teach you the fling saw more design iteration than any other part of the game, so you can bet that they’ll be streamlined and full of clever psychology. In this article I’ll be looking closely at the series of levels that slowly build up the fling, and after that I’ll look at the deeper strategic variations on it that are presented.
In the following rooms, the player is able to control the position of the blue portal only, the orange one being set down immovably by the designers, a limitation that is used to make your education smoother. The first part of this education is a room that is almost insultingly simple to “solve”:
All you have to do here is put a portal somewhere and go through it. It would be wrong to call this a puzzle, really. It’s an antepiece, a brief task that exists not to provide challenge in itself but to introduce an idea that will help the player deal with a setpiece that will come soon after it. Another example of an antepiece in Portal is the incineration of your companion cube, which gives you a method you will later use to kill the last boss.
The action introduced here is a forward movement, out of a portal, across a gap, to a place you couldn't reach (the player will have found that that ledge is slightly too high to jump onto). This action also becomes associated with the very visible thrust-forward panel with the orange portal on it, which will become a motif.
Straight away there’s the thrust-forward panel again. The clever thing that can happen here, which I've tried to depict in the gif, is that the player is lured into making a mistake that will actually bring them closer to a solution.
It may be a surprising concept to a player that you’re supposed to fall into a pit. So note that the platform over the pit is shaped like a diving board. This encourages the player to dive off it, and if it isn't enough, the player is likely to fall off it anyway because it’s so narrow.
When you’re in the pit, there’s not much you can do other than try to get yourself out, and since this pit’s walls lack the white texture that other pits usually have, you have to put a portal on the floor in order to get out. So you do that, and you “plop out” (emerge with low momentum) of the orange portal panel, to a point near to the diving board again. And then there we are: your portals are now set for you to solve the puzzle, or at least give you a clue. You dive in again, and the momentum you build in falling takes you to the exit.
It is actually possible to have blundered through that previous room without even having looked directly at the thrusting panel with the orange portal on it, and therefore without understanding how you solved it. This room makes sure you get a better idea of what’s going on.
The designers have put the orange portal in the pit, so to prepare for your fling you must place your portal on the panel above - so you’ll have engaged with both parts of the set up. Famously, playtesters would take ages to look up and see the white surface they need, so to draw attention to the panels the designers gave them protrusion animations. They’re loud and slow-moving, so you’re unlikely to miss them.
The player flings themselves twice in this room, and all the variables (fall height, gap length, elevation of destination) get changed. This is a step toward the player being able to see the system and manipulate those variables themselves. Bearing in mind that the fling is so fast and disorientating (the second fling turns you upside-down!), it’s nice to be able to gain this understanding at a slow pace. As you leave the room, the voice over remarks “momentum is conserved between portals”. The player can link this neat fact to intuition quite easily.
Between the chamber just described and chamber 12, we’re given control and responsibility for both portals. Chamber 12, then, exists mostly to make you handle both ends of a fling.
You start at the bottom, and you ascend by placing “fling-out” portals on the panels, to fling to places you won’t actually be able to see. When you’ve gotten to the top, you’ll be in a situation where your “drop in” portal is quite far away and out of sight of your “fling out” portal. This means you’re now comfortably using resources in two separated places. You’ll also fling from a skewed panel here, which is powerful and novel.
There’s an interesting, well-chosen structure here. An enjoyable vertically-progressing level is a rare beast in first-person games, and that is how you move through here, with flings out of successive panels as they slowly become visible to you. But then you keep returning to the bottom because that’s where the “fling pit” you’re using is (you even make a return from the very top shelf where the exit is, because you have to bring a cube down to a button). That means that you can understand and grow accustomed to these odd, initially-obfuscated spaces as they unfold, giving you a real sense of the privilege of having both portals and using them to explore lofty locations.
We’ve taken a big step forward in the game. This room uses a thrust-forward panel to reminds you of the fling; but in contrast to previous areas there’s no pit to jump into. Understanding the fling means knowing the difference between exiting a portal with momentum and just plopping out of it. Without your pit to build momentum you can only plop.
The elegant solution is to use your plop as a descent. So the solution involves going into the same portal twice in a row, but with different results. Since flings are disorientating, players can have difficulty realizing they need to plan this, but when you get it the movement feels lovely.
There’s an aesthetically pleasing clue in the layout: the first half of the pre-wall area is white, while in the second half the walls and ceiling are black. The place on the floor you “plop” onto is at this precisely marked half-way point. The designers are suggesting to the player who arrives here “you have done half the job, you may want to double up what you just did”.
The puzzle here very strictly forces you to place a portal while falling through the air, such that you will go straight into it - not so easy, since you’re emerging from a fling that has rotated the direction you perceive as “up” by 90 degrees!
Here, if you plop out of the portal on the thrusted panel with zero momentum, you’ll just fall onto the black unportalable floor. That’s why you use the petite fall in the room on the right to get a minor fling - not enough for you to clear the glass barrier, but enough to get you to the white, portalable floor. It is when you’re falling towards that surface that you slap down a portal in front of you, allowing you to chain directly into a second, more powerful forward fling that completes the puzzle.
You have to climb some stairs to get up to the small fall. This is possibly to emphasize the fact that you are getting some usable elevation here, small as it is.
This challenge puts the last parts of your tool in place. You can make a plan involving a descent divided into two quantitatively different sections, in two separate rooms. You’re trained in repositioning a portal while in the air, and you’ll probably find that you’re capable of making the decision to do that on-the-fly (lol). You will think nothing of reappropriating the fling as a convenient means of transportation - which is sadly what you are required to repeat ad nauseam in Portal’s “behind the scenes” sequence, and Portal 2’s.
Strategic extensions of the fling
Now that we have a thorough understanding of the fling, we can create and enjoy interesting puzzle solutions by putting mathematically natural spins on it.
This is like the previous fling, but you’re falling towards a portal on a miniscule piece of floorspace. It takes a delicate jump (from a portal in the hallway) and air control to to hit the target.
This puzzle from the “Advanced” version of chamber 15 is also more about dexterity than brains, but it’s a very cool kind of dexterity. You have to get a portal on the thrusted panel while doing an endless fall between portals on the white squares. This requires the shooting of a target that is (from your point of view) rapidly cycling upwards.
What happens if you do a fling out of a portal that’s on the floor? You fly high into the air, and then fall back down, into your portal - and then you come out of your other portal, and repeat!
When you’re flung out of one of the platforms in this level, at the apex of your trajectory you have a chance to slap a portal on the surface of the next platform up. You climb these mammoth stairs by repeatedly doing that - and remembering to alternate firing blue and orange portals! Note that the platforms resemble the earlier thrust-forward panels.
Your body does a complete sideways roll whenever you pass through your upward-facing portals. Initially people try to counter this movement, trying to get a clear shot by staying upright, but eventually they learn that it’s better to just let the movement happen and try to aim as soon as you’re upright. It’s an amusing feeling to try to compose oneself while falling at terminal velocity and periodically doing side flips. Fortunately you never lose any momentum, so you can go through a “portal-trampoline” on a given platform as many times as you like before taking a shot at the next one.
Putting two portals on the floor is one of the first experiments a lot of people do when they get the gun, and they find that it often just results in a mess. If you did that experiment, this is a rewarding chamber to come to. There’s a game design principle to be gleaned from this: we must look for the things that people tend to enjoy doing in our engines, and then make levels with some practical excuse for doing those things.
The above puzzle is by far the best piece of design in the whole of the “Portal: The Flash Version mappack”. Here the portal that you initially go into is the same as the portal that you eventually fling out of, so there’s that cute fall on the left side of the picture where you come out of the blue portal and then immediately reposition it such that you fall directly back into it. It may seem simple, but realizing you can conserve that orange portal where it is can take a lot of thinking - then when you get it, there’s a delightful symmetry to it.
In the real Portal there’s a puzzle in the Advanced version of test chamber 15 which has the same solution. Those of you who have played it will see that the sloped ceiling that you bounce off at the end here is a very conscientious touch. Valve screwed up their version of this somewhat; instead of the bounce, it’s like you have to manually air control yourself onto the ledge, which is a fussy dismount from an elegant routine.
Finally, this article would not be complete without a picture of this, the final puzzle of Narbacular Drop, the student game that would eventually become Portal. In the middle of the screen is the avatar, princess Noknees; the two monster-faces you can see are the portals.
This was the first proto-fling ever to be implemented. It’s situated at the top of a tall tower which you climb mostly by positioning portals over platforms at precarious angles. The room as a whole has something in common with chamber 12, though less like a piece of shapely Rococo furniture, more like a multi-storey car park.
In theory it’s cool to have you climb almost all the way up a tower in one way, and then have you ascend the final step in a novel but somewhat-related way. However, I think that surprising puzzle solutions like the fling require focused puzzle design, so to me this is very ugly. It’s certainly enclosed and obfuscating. When you get the solution it doesn’t feel so smooth or substantial. And chances are that you won’t get that solution without having wasted some time searching for clues or getting fixated on unimportant details in this level. The time-wastiness is compounded by the room being deadly and unforgiving, and therefore discouraging experimentation.
When people talk about works of interactive entertainment aspiring to offer some insight into the world, they tend to advocate titles like “The Walking Dead” and “Dear Esther”. That is sad to me, because it seems to suggest that to be artistically insightful, a game must be strategically dull.
I feel that the most beautiful pieces of interactive entertainment are the ones that communicate through strategic decisions you make. This means that the things you communicate are mostly “mathematical”, so I usually use comparisons with Bach and Islamic art when I’m trying to explain to people the kinds of artistic insights video games can offer.
I believe Portal contains quite a lot of mathematical communication. I would not claim that the puzzles teach you anything about the wormholes that may actually exist, with the possible exception of the ones that focus on the energy pellet. But it seems to me that the solutions I’ve discussed here do have a thing or three to say about the real world. For example: when real objects are moving at high speed and there is a sudden change in the direction of their acceleration, they move through a curve that you will be able to recognize, as a person who has experienced the “fling”. And that may not be the kind of insight that you usually expect artists to work on, but it is communicated beautifully, which is why we need even more good developers working on things like Portal.
Portal offers a brave new philosophy of puzzle design: “we value puzzles not because we like struggling, but because we like having mechanical phenomena revealed to us”. Less like math questions, more like interactive math papers.
When you’re writing a math paper, you want to be as clear and concise as possible, which is what we have in Portal. A designer plays around in a game engine, discovers some naturally emerging phenomenon like the fling, and says “now to design a puzzle which involves the player experiencing this phenomenon”. In designing the puzzle, to paraphrase Einstein, we must make things as simple as possible, but no simpler. No red herrings; no unnecessary extra tasks or things for the player to get distracted by. Use stepping-stone puzzles to make the player’s discoveries active, concise, and thoroughly-established, and only allow things to be repeated if there’s a real chance the player hasn't understood something yet. This is a philosophy you only see in post-Portal puzzle games, with a few wonderful exceptions.