Video Game Replay: Portal/Portal 2
Spoilers Ahoy
No seriously, I’m about to spoil two of the best games of the last 20 years, and if you somehow still haven’t played them, bookmark this post and head over to Steam right now trust me.
I’m Serious, go play it.
My kids had never played either of the Portal games, so on a whim a couple of weeks ago we fired them up on the SteamDeck and played through them as a team. (Technical sidebar: the PS5 controller makes an excellent bluetooth controller for the SteamDeck when it’s connected to a TV, and really easy to set up! Ironically, a million times easier than trying to use my old Steam Controller.) I played them both when they came out, but hadn’t since.
Portal is a perfectly crafted jewel of a game. The gameplay is perfect, the puzzles are interesting, the design and look of the game perfectly matched with what the game engine can do.
It’s also got maybe my all time favorite piece of narrative slight-of-hand I’ve ever seen in a video game.
Recall that the frame for the game is that you’re Chell, a “test subject” for Aperture Science Labs, testing out their “Portal Gun.” Structurally, you move through a series of levels, each of which is a confined space where you need to use the gun in increasingly complex ways to make portals to get from the entrance door to the exit. The portals themselves are person-sized wormholes or connections that you can drop onto most flat surfaces, connecting disparate areas of the geography. But also, objects—including yourself—keep their momentum as they pass through the portals, so not only can you use them to navigate around obstacles but to build a variety of slingshots, catapults, launchers. You redirect lasers, confuse turrets, bounce objects. Critically, you also don’t have another kind of gun, just the portal one, so puzzles that in a “regular” first person shooter would be solved via firepower here have to be solved by variable cartography.
The puzzles are from the “duplicate, then elaborate”school of design, each one adds some new twist or obstacle or complication that you have to combine with what you leaned last time.
The only other character is the robot voice that’s giving you instructions—that’s GLaDOS, voiced by the staggeringly good Ellen McLain, who seems to be running the show. She’s a computer mastermind in the HAL/SHODAN sense, but a little ruder, a little funnier.
Each test chamber has an opening graphic or placard, giving the chamber number, counting up to 19. The opening sign also has a series of icons indicating which obstacles this room has, with the array lighting up more and more as you move through the game.
The visual design of the game also perfectly matched what the upgraded Half-Life 2 engine it was using could do. The test chambers were mostly white high-tech spaces, sort of 2001 crossed with the Apple store, with the occasional moving panel or window. Big doors slide open to reveal pneumatic tube–like elevators between levels. Metalic panels indicate walls that can’t have portals opened on them, as opposed to the normal glowing white walls. Most of all, the visual design was very clear and focused. Considering the strange geometries you could create with the portals, this was critical to making the puzzles solvable, you could always get your bearings and get an eye-line to where the exit door was, regardless of if you could see how to get there yet.
This is where I pause and remind everyone that Portal wasn’t released on it’s own. It was the “other, other” new game in the Orange Box collection, bundled with Half-Life 2: Episode Two and Team Fortress 2. Portal was clearly the one they had the least commercial expectations for; Team Fortress got all the ads and early chatter, Episode 2 was exciting because it was moving the Half-Life story forward, Portal had the quality that it was the bonus track on the album, the fun tech demo.
And so there was no reason to believe that Portal was anything other than it presented itself as: 19 puzzles with this cool portal tech, which would presumably show up in Half-Life 3 as part of a “real game.”
If you paid attention, though, there were some indications that things weren’t quite right. Every test chamber had at least one observation window looking down into it, and while you could see chairs and computers, you never saw a person moving around on the other side of the translucent glass. GLaDOS wasn’t ever openly malevolent, but sometimes seemed a little off. And there were a few places where you could slip “backstage” of a test chamber, and find strange graffiti and other abandoned debris. There was nothing you could do to interact with it, though? GLaDOS never mentions it? Just a fun little easter egg, I guess, like the G-Man peeking through windows at you at the start of the first Half-Life A little strange though, for a glorified tech demo?
So then, when you get to Test Chamber 19 and then instead of the game ending GlaDOS tries to dump you into the incinerator, you get to have the absolutely breathtaking realization that no, you fell for it, you didn’t just beat the game, you beat the tutorial.
The rest of the game is making your way through the infrastructure of the testing facility towards GLaDOS, using all the portal tricks the game carefully tought you earlier. You find out that, hey, the reason you never saw anyone behind those windows was because GLaDOS killed them all, and now instead of a fun tech demo puzzle game you’re in a 1:1 duel to the death with an evil computer. It’s great! Then there’s a song at the end!
Part what makes it so great is the length: it’s not short short, but it knows how not to wear out its welcome. Replaying it, I think we beat in in three after-school nights, neither rushing nor going terribly slowly. Perfectly paced, satisfying without being overlong, trim without leaving you feeling cheated.
It did, however, leave everyone wanting more.
It was, and I’m marking it down here, a huge success. Portal ripped through the circa 2008 nerd culture like few things I’ve ever seen before or since. It quickly flipped from “the bonus track” to “really, there’s no way to get this without that dumb-looking Team Fortress?” The cake memes were everywhere. Making a sequel was an absolute no-brainer.
They announced Portal 2 in 2010, it was released the next year. Unlike the first game, this was a full triple-A standalone release. In a world where it had already become clear that Half-Life 3 was never going to happen, this was Valve’s Next Big Thing. Structurally, Portal wasn’t a lot like Valve’s other work, Portal 2, on the other hand, was absolutely A Valve Game.TM
This is where I pause and admit that my opinion most of-of-step with the video game–playing mainstream is that I do not, personally, care for either of the Half-Life games. This is not a contrarian hot take, I’m not about to try to convince you that they’re Bad Actually, I understand why they are as popular and beloved as they are, I am aware of all the ways they were incredibly innovative and influential.
I feel the same way about the Half-Lifes that I do about Cola: I acknowledge that it’s very popular, don’t have anything against it, but it is not my preferred flavor. I guess, in this strained metaphor, the original Deus Ex is Mountain Dew?
Because this is going to be relevant in a moment, let me attempt to sketch for you what I don’t like about them. I’ve thought about this a lot, because it’s very strange to beat a game, think to yourself “well, that was okay I guess, but not that great” and then have everyone you know declare it to be the greatest game of all time, and then have that happen even more so with the sequel. You gotta stop and make sure you’re not the idiot, you know?
Valve shooters tend to be extremely linear games where you make your way though an environment, alternating segments of “traversal” where you have to find the one way forward, and “encounters” which are either an in-engine cutscene, a shootout, or more rarely, a puzzle to get past. They very much like to imply a larger, more complex environment out and around you, but all the doors are locked and impassable except the one door or vent you can go through. It’s all stage scenery, basically. And while it’s cool that the cutscenes don’t take your control away, it sometimes feels like you’re watching the game get played for you. In my less charitable moods, I describe the Half-Lifes as “slowly walking down an elaborately decorated single hallway.”
And the obvious follow-up question here is, well buddy, even just limiting ourselves to first person shooters from the turn of the century, that also pretty much describes Max Payne, which you loved, so what gives? Broadly, I think it’s two things. First, those fake environments. I prefer sprawling non-linear environments in games, but I don’t mind something more linear. What drove me crazy about Half-Life 2 especially was you’d get these vast city-scapes, and then only a tiny little alleyway was available to you. Vice City had already been out for two years! Deus Ex did all kinds of things with open spaces on limited computers! Max Payne didn’t irritate me as much because you spent all your time in naturally-enclosed areas; abandoned subways, empty office buildings, and the like. I spent a lot of time wishing City 17 was more like Hong Kong in Deus Ex and less like the Black Mesa facility.
But mostly what I didn’t like was I thought most of the actual shooting was pretty boring. I like games that structure “encounters” more like puzzles—this is why I prefer turn-based tactical fights in RPGs, why I like X-COM more than Diablo, and so on. One of the things I loved so much about Max Payne, was that between the fact you really could take cover and the bullet time mechanic, each shootout functioned as a puzzle—how do I get through this without being hit? More than once I’d get through a fight, and the reload, muttering “I can do better.”
The parts of Half-Life 2 I really liked—the sawblades vs zombies village, that big physics puzzle with the crane—were encounters that functioned more like puzzles. It wasn’t just “keep an eye on your ammo remaining and watch the floor for those crab things.”
I disliked the way Half-Life 2 would get you to the next set-piece, and then say “okay, this is a gravity gun puzzle” or “nope, this is just shooting,” or “yeah, this is a laser-guided missile puzzle.” There were very very few opportunities to mix and match, or find your own solution to anything.
This sounds like snark but isn’t: my favorite part of Half-Life 2 was the final level where you have to use the gravity gun to bounce those energy spheres around and disintegrate things. That was something new, and didn’t play like anything else. I wish the whole game has been like that.
I bring all this up because Portal 2 has this exact structure, and I loved it.
Portal 2 opens with the swagger of a game being make by people who know they’re making a hit. Portal sometimes has a slightly hesitant quality to it, beyond just being the “bonus game,” in that you can tell the developers aren’t quite sure if the audience is going to buy what they’re selling. Portal 2, on the other hand, is clearly made by people who know the audience loved what they did last time. It has a really solid take on what worked from the first game and leans into them. Among other things, that means more humor and more atmospherics. It also knows it has more space, so it settles in, puts its feet up, and gets comfortable.
Valve hadn’t been known for funny games, and while Portal was funny that humor tended to be subtle and deadpan. But the jokes were everyone’s favorite part, so Portal 2 comes out of the gate making it clear that this is a comedy: a terribly dark comedy, but a comedy.
It opens with a fairly bravura set-piece, where you start in what looks like a 1950s hotel room, do a couple of tutorial moves to learn the controls, go to sleep, and then wake up terribly far in the future. The room is ruined and overgrown, and things have clearly gone wrong. The first new character of the game, Wheatley, quickly arrives to finish your tutorial. He’s a spherical robot driving around on a track on the ceiling, and he’s played by Steven Merchant, who at the time was mostly known for the UK version of The Office. The opening turns into something of a technical flex as Wheatley starts driving your hotel room around on a larger set of tracks, crashing into things, disintegrating the walls, as you have to move around and avoid being thrown out. As the walls fall apart, you get glimpses of that same backstage infrastructure from the first game—you’re still in the same Aperture Science facility, just in a new part. On paper, this is a classic Valve “live action cutscene”, a lot like the opening train rides of both Half-Lifes, but the key difference for me was that it was very funny. The slapstick of the room crashing into things, Wheatley’s stuttered apologies, great stuff.
You’re once again playing Chell, a silent protagonist in the style of Half-Life’s Gordon Freeman. Unlike Half-Life which dances around why Freeman never says anything, here’s it’s lampshaded directly; Wheatley thinks you have brain damage, GLaDOS later refers to you as a “mute lunatic”; the writer, Erik Wolpaw has said several times that she just refuses to give anyone the satisfaction of a response.
The utilitarian, 2001-esque test chambers of Portal were very spooky in their own subtle way, and then the backstage areas even more so. Portal 2 knows not to try to recreate either of those, but keeps finding new ways to riff on the same basic environmental grammar.
You quickly find yourself back in the facility from the first game, but long-abandoned and gone to ruin. The first few levels are the same intro test chambers from the first game, but now overgrown and abandoned. It’s an inspired way to reacclimatize returning players to the game while also onboarding new ones, while still making it clear this this game is going to be different, and very spooky.
But, like the first game, Portal 2 knows not to overstay its welcome with any particular batch of ideas. The game passes through, roughly, five acts. After the opening act in the ruined facility, you accidentally wake GLaDOS up, and she retakes control, and she decides to get back to work.
This second act is the one most the first game, with GLaDOS running you through new test chambers. The facility itself becomes much more of a character, with the chambers “waking up”, walls reorganizing themselves, the various panels shaking off years of debris before re-assuming their test configurations, becoming less ruined and more like they were before.
The best example of the second game’s swagger is the way it uses GLaDOS herself. While she was used sparingly before, here they know she’s the best part of the game, and make sure to use her to the fullest. Her voice is less artificial, and she has more things to say, and they’re funner.
My favorite example of this is that as her frustration mounts, we end up with an extended series of jokes where rather than questioning your skills or value, she just starts calling you fat in increasingly bitchy ways. GLaDOS is far more human in this game to the character’s immense benefit, there’s a sense that her behavior in the first game is her “professional demeanor”, and in the second game she’s gotten tired and frustrated enough that the “real her” is spilling out.
While this is going on, most levels have a spot where Wheatley peeks through a half-opened panel or around a corner. A carefully-designed set of blink-or-you’ll-miss-it encounters that make sure you never blink. Eventually he stages a rescue, and the third act is once again backstage of the testing facility, making your way towards GLaDOS. Similar in design to the backstage second half of the first game, the facility here come across as larger and more menacing, with more things going on that just your strange tests. Views recede into a blue haze past the industrial strutures, where is all this, exactly?
The closest the game comes to replicating the first game’s surprise twist is at the fight with GLaDOS—it looks like so far we’ve mostly been re-staging the plot of the first game with better graphics and funnier writing, but then Wheatley takes over, goes all megalomaniacal, straps GLaDOS to a potato battery, and throws the pair of you down a long shaft.
The best, and most famous part of the game is the fourth act, set in the abandoned 50s, 70s, and 80s–era testing facilities. Turns out the whole facility was built inside an abandoned salt mine, working from the bottom up, and everything you’ve seen so far was just the very top layer.
This is where we meet the last new character—Cave Johnson, played by JK Simmons in full “bring me pictures of Spider-man” mode, the founder and now deceased CEO of Aperture Science, via his leftover recordings. Johnson’s rants, and GLaDOS’s snark in return from her position as a potato perched on your gun, makes for the game’s best writing.
This is where the game most settles into it’s Half-Life 2 style structure, you alternate between navigating your way up to the next level through the abandoned structures, then solve a test chamber or two designed with an appropriately retro style of tech, and then go back to traversal. Like the first game, it does a remarkable job of teaching you some new portal tricks with the test chambers, and then letting you loose to use them as you try and move around between those test chambers.
It’s worth noting how much exposition they cram into the jokes Cave Johnson and GLaDOS make at each other—most specifically how much time they spend talking about moon dust, which seems like just another wacky detail until you find out why, and realize they’ve been giving you the solution to a puzzle the whole time.
Finally, you make it back up to the “modern day”, facility, where things have gone horribly wrong with Wheatley in charge. It’s a remarkable piece of design work that, using the same basic pieces, the freshly re-ruined facility manages to be the most menacing yet. It’s positively apocalyptic with tangled up rooms and looming fires on the horizon as you try to keep the whole place from being destroyed and solve Wheatley’s terrible puzzles.
The key difference structurally between the two games is that the second knows it can’t recreate the Big Surprise of the first, so it doesn’t try. Instead, the second game is built around anticipation, each act has an end goal that gets declared at the start and that you spend the whole time working towards: escape the facility, escape GLaDOS, climb back out, defeat Wheatley. While this keeps the game moving forward, it does tend to blunt the puzzles a little; unlike the first game there’s a tendency to try and rush through them so you can see what happens next.
That’s part of how Half-Life 2 structure’s worked too: you’d get a goal, then fight your way through whatever it was to get where the goal needed you to be.
Which brings me back around to why did I like Portal 2 so much more than the Half-Lifes? For starters, I like the humor a lot more than the post-apocalyptic melodrama. Mainly, though, it’s the puzzles. While I found the shooting encounters frequently boring, the portal puzzles never were, and kept building on themselves in fun and interesting ways. There was never an “oh this again” moment, there was always some new twist or “yes and”. And whereas the linear and confined nature of the Half-Lifes felt limiting, here it made the puzzles feel even possible. Knowing there’s one way through keeps the tangled wreckage at the bottom of the test shaft from feeling overwhelming. You’re not going to get lost, you’re not going to chase the wrong path, let’s just look around for the one place you can shoot a portal and keep moving.
As an aside on that point: there’s a regular Discourse that pops up with video games around how much player affordance is too much, every 9–18 months someone would get mad about yellow paint on ladders back on the old twitter. Portal 2 does a really elegant job of this by using light; most of the facilities are very dark, especially the older ones, and the few spotlights that are there will just casually play across the area where you need to shoot a portal. It’s a slick way to draw the eye without making it insultingly obvious. (There are a few places where you’d have a collapsed bridge but then the fallen wreckage would just happen to form a perfect walkway over to where you need to be, which gets a little eye-rolling.)
Both Portal games are a masterclass in this, in game design that subtly wiggles its eyebrows at the right answer and then lets you think you solved it all on your own.
Narratively, the game has a pretty conclusive end, there’s room for more but no real un-pulled threads. From a design perspective, this also felt like the definitive statement on these mechanics. Half-Life 3 has become a vaporware meme because there’s still so much plot and mechanics you could build on top of those games, but conversely no one really clamors for a Portal 3, because it doesn’t need one. Any new game with those portal mechanics would need to do something new, something different, and whatever that might be, it wouldn’t be Portal. The Portal/Portal 2 diptych might be the only perfect 1-2 punch in all of video games, and there’s no reason to make more. Outstanding work, just as fun over a decade later as they were when they were new. I’d say something like “they don’t make ‘em like that anymore,” but no, they never made them like that at any time, except those two.
I will just throw this out here though: I’d pay real money for a game just called “Three” that let you play as Gordon Freeman, Chell, and Alyx simultaneously, swapping between them to solve portal/gravity/bullet gun puzzles as you had to team up with GLaDOS to defeat those aliens.