Portal: Alive & Kicking Escape Design Post-Mortem

Posted 2026 Aug 28

This is a reflection of the several years on and off that I spent designing the Escape sequence for Portal: Alive & Kicking. I've found the response to it be pretty divisive, so I thought I would talk about the approach I took and the problems I had to solve. I may adapt this into a video later.

How do you port Portal 1's geometry into Portal 2? The two games have vastly different ideas on how the facility is structured. In terms of detailing, Portal 1 feels like crawling around the cramped guts of a machine while Portal 2 is much more spacious and draws more on a fear of heights. Portal 1 Escape gameplay is rooms, while Portal 2 BTS gameplay is catwalks and exposition. (There are ofc counter-examples, especially in the gameplay moments leading up to the final boss.) Porting the original level geometry directly was, in my mind, not a viable option. Instead, I wanted to capture the feeling of crawling around in the guts while keeping more faithful to the kind of architecture that Portal 2 demands.

The end result is that I ended up using a lot of concrete and not a lot of test chambers. It comes off a bit "old Aperture" since that's where those textures mostly get used in Portal 2, but I tried to make sure that there was modern equipment everywhere, as if this was maybe an old part of the facility that had been retrofitted with new infrastructure. Is that "canon"? I don't know. I think it worked out OK. I don't love every part of the artpass, but I've gotten a lot of feedback that it captures the vibe of the original pretty well.

Another part of capturing the vibe of the original is that I personally found the original game's layout for Escape to be intentionally directionless and/or disorienting. On reflection, I think this only applies for certain portions of the escape, and other parts are much more straightforward. In any case, I leaned into the idea that the facility is uncaring and you are not meant to be here, and that this feeling should be part of the gameplay and not just the artpass. Many portal transitions are intentionally awkward, and the path forward is not always entirely clear.

There's a common debate in game design about friction - Should games be streamlined to avoid confusion? Is it ok to leave a player confused or frustrated? How much streamlining is too much or too little? Every designer is going to have a different answer. However, I have found that level design as a discipline seems to lean heavily in favor of streamlining.

I don't completely agree. My thought on the matter is that the point of streamlining or avoiding friction in some aspect is a tool for getting out of the way for some other element of the game to shine through. In most cases, the gameplay is the star of the show, and the level design exists to facilitate that. This is where it makes a lot of sense to obey rules about "guiding the player" with lighting, detailing, and architectural cues to get them where they need to go. I think this has resulted in the idea that streamlining level design is *always* a good idea, and that's the part I disagree with. I think navigation as a puzzle is a perfectly valid gameplay moment to insert into a level.

That all said, I do not think I did a great job with A&K in this regard for a few different reasons. One is that, at the time that I was designing a lot of these maps, I was working on my architecture degree and/or working in the architecture industry. This comes through in my level design in terms of making maps that feel structurally or formally realistic. However, this also became a bit of a crutch. A common theme among the escape levels, especially 00 and 01, was this idea of fitting everything into a rigid structural grid. This meant a lot of layout decisions were decided more by the grid and less by the desire to create Good Gameplay. While I still believe in the importance of structural realism in most forms of level design, I have since gotten better about establishing structural systems that don't get in the way of Good Gameplay.

In addition, I really struggled with the lighting for basically the entire escape sequence. I think the end result is decent, but there's way too much darkness. The entire sequence needs a brightening, and I will probably update this in the future. This pervasive darkness makes gameplay notably worse - if the intended gameplay moment of a space is a navigation puzzle, it would help if the player could visually process everything more easily. If the puzzle is navigation, who is to say that the route isn't somewhere in the darkness? This is something Portal 1 does pretty well - most of the escape sequence is lit very well.

Lastly, I think this idea of navigation as a puzzle was severely over-applied. There are a decent number of rooms where figuring out where to go is an intentional puzzle, but it is paired with another puzzle at the same time, and the two do not agree with each other. If the puzzle of a room is to figure out how to clamber along some pistons to the exit, it would help if the exit was easily identifiable. One layer of friction is fine, but two layers of friction tends to be too much. As an example, the "piston jump" room has a row of turbines separating the main chamber from the pistons, and it's a layer of obfuscation that really just isn't necessary when the piston jump mechanic is already interesting on its own.

Making this issue worse is the lack of backtrack stoppers. Valve likes to frequently insert drop-downs or other methods of preventing players from backtracking, which helps the player understand what parts of the map are currently "in play". I did not do this frequently enough. I think that for a navigation puzzle to work, you need to limit the player's options.

I would estimate that I made enough content to fill at least three escape sequences. It took a lot of work to develop my thoughts on what the escape sequence wanted to look and play like, and a lot of work to actually execute on it (or more accurately, to the develop the skills necessary to execute it). After 10 years of working on and off on this thing, I just needed to move onto something else if I wanted to actually develop my skills more instead of digging myself a deeper hole. I think I can speak for the entire team when I say that we were all tired of me redoing the maps over and over. We just needed to ship the damn thing.

Some thoughts on the individual maps:

19bts: I like how the start of this turned out - crawl around on the roof of the chamber, get more comfortable with which surfaces are and aren't portalable, and tease the player a bit with an obscured look into an office and some closed doors into observation rooms. I think this section is a good navigation puzzle - it's a bit awkward, but there's nothing else going on. I think it communicates the feeling that you shouldn't be here.

After this is the double fan room. This introduces the player to some of the more awkward portal transitions that they're going to be encountering in the future. I'm not sure how I feel about the vent immediately after this - players don't seem to struggle with it too much.

There's the more open room after this, I'm not really happy with how this bit turned out - the Valve version of this room allows a few different approaches to get onto the catwalk, while the A&K version allows only one approach (a fling). I think the Valve version does a better job of conveying the feeling of being an intruder.

Then there's a little fling puzzle that I think turned out ok (although it could have benefitted from a backtrack gate), the tube ride, the test chamber, and a little ending segment. Fun fact: I actually wanted to remove the tube ride because we couldn't find a way to make it not feel kind of janky. I got pretty much unanimous feedback that the tube ride absolutely needed to stay, so the jank stayed too. Another fun fact: the last part of the level immediately after the test chamber was designed by Dot Operator because we needed more space to let Glados say her lines. We took the opportunity to introduce flipping a switch to active a piston.

00: This is probably my least favorite map of the bunch. The pistons are already enough of of a gameplay moment that the navigation should not have been confusing. This is what I mean when I talk about multiple layers of friction - there's too much going on and it can lead to a frustrating friction rather than an engaging friction. I redid the greybox for this level several times and what you see is the best form of whatever was going through my head at the time. Also, the lighting is way too dark.

That said, I do really like the piston jump idea. Why did I stick a line of turbines in-between the pistons and the main chamber? I don't know. I kind of hate it.

01: This is probably the most significantly different part of the escape as compared to the original game, which was because I wanted to introduce the player to claws, since one appears in the final boss fight. I like the general idea of this area, but it was bogged down by my strict adherence to a tight grid, as well as those overlapping layers of friction that I keep talking about. In general I still like the turret tunnels, but they could have been executed much better.

The rocket turret lab portion of the level is probably the part of the game that I think holds up the best. In an earlier draft of this map, we had a battery fetching puzzle where the player would need to fetch a number of batteries from different small areas, insert them into a machine at the hub, and then progress. The point of this moment was to keep the rocket turret in the hub so that the player does not need to keep walking back over and over to get to the turret again. We ultimately ended up ditching this after a *very* long discussion, mostly because it felt strange to introduce batteries as a mechanic and then never use them again.

The version you see now is my version of solving that issue while keeping the layout linear. The layout loops back around itself so that you get a view into the rocket turret room from above. A simple and elegant solution that isn't anything amazing but took a while to arrive upon.

Also of note is that there is no moment where you blow up a tube. I don't know if this was a common experience, but I got completely stuck on that part in my first playthrough of Portal 1. Because of that, I replaced it with blowing up an observation room window, which I think is more intuitive while still getting that feeling of Breaking The Rules.

02: Boy, this went through a lot of iteration, and the final result definitely has some issues. Let's talk about the ambush first.

I like the ambush. It is meant to feel messy - you probably aren't getting through without getting shot a little bit. That said, it doesn't seem to occur to a lot of players to crouch behind the plywood cover provided by each turret. I think this could be improved. Also, a lot of players seem to struggle with getting to the turret with no nearby portalable surfaces. This turret exists to be the last turret the player topples, which then (hopefully) gets the player in a good position to see the relevant portalable surfaces for continuing on.

The sequence after this is messy in a way that I don't like. At one point this segment had you jumping from the 1st floor to catapult up to the 2nd floor, and then performing the same jump into the pit to catapult even higher up to the 3rd floor. This actually required getting more momentum than Portal 2's terminal velocity allows, so there was a complex setup of catapults to make it all work. It was too janky to keep, so a different solution was retrofitted and I don't think it works very well. Turrets on claws are brought back again to remind the player about what they'll find at the boss fight.

A fun fact about the intro segment - the ambush arena originally had deadly goo in the flocculator pit, but we had a very mysterious crash related to this specific room and placing portals in specific areas. Getting rid of the goo made the crash go away. This prompted me to completely redo the rooms leading up to the ambush so that no goo would be rendered from that room.

03: I don't have a lot to say about the boss fight design. I redesigned it to use the claw, and everything kind of naturally followed from that. That said, I do have some fun anecdotes.

One: The third claw puzzle used to be a game of keep-away. It was really fucking annoying, so I had it just go around in a circle. It's still kind of annoying, but I couldn't come up with anything better to escalate from the previous puzzle.

Two: The entity setup for making the core launching work is a nightmare. The claw itself is controlled by a vscript, and cores are launched towards the claw using a catapult. However, there was a fundamental problem - the cores are wedged into Glados' collision model, so they often get thrown off-course. The easy solution to this is to disable Glados' hitbox for a moment, launch the core, and then turn it back on. However, I couldn't find an entity that allowed toggling the collisions while also having an output for taking damage from rockets. Maybe there is one! Not sure. What I ended up doing was giving Glados a brushwork collision model and yeeting it into the ceiling before every core launch. That said, I'm not sure this was necessary at all, because of what I discovered next...

Three: Sometimes the core collides with the screens around Glados. Sometimes the core is slightly off target and misses the claw entirely. What do we do about that? What actually happens is that there is a simplified invisible duplicate of the arena somewhere else in the map where the same core fling is performed without any obstacles in the way. The core you see in the map is puppeted by its duplicate, using a vscript to parent the position but not the rotation. If even all of that fails, there's also a bunch of triggers surrounding the arena and the floor that simply teleport the core to the claw no matter what.

Four: The day before release, I encountered a bug where Glados simply decided to stop talking. We were not able to replicate it. No idea what happened, but I guess that's just a thing.

Boy, that's a lot of words! Hopefully it was worth reading. I'm far from proud of everything we shipped, but overall I think we made a good game. Not great, but good. I want to move on and do something else because this has occupied my brain for a decade and a lot of the emotions I associate with it are not necessarily positive.

Speaking of moving on, did you know I made my own puzzle game? It's called Clover's Quadrants and I think it is significantly better designed than this mod. It also only took me 2 1/2 years to make! Wow! Please buy my game.