That PC superdeveloper Valve hired a group of students to work on their own title was surprising enough news, but no-one could have imagined that the game they would make would be many people's pick of 2007. We caught up with Kim Swift to talk portals, education and companion cubes.
Kim Swift was level designer and team leader on Portal, and prior to that a student at Digipen Institute of Technology, where she and her student team-mates were discovered by Valve.
It would appear that some games studios are cautious about hiring graduates – do you see the Portal story as being useful against that?
The best part of our degree was that we had to make games - we made a game every year from the time we start. Each time the requirements are a little different - the first time was a text-based game actually, and then by the time we graduate we have to have a full 3D game with simulated physics etc.
What I think developers are edgy about is looking at someone's portfolio and seeing that they've done a thing here and another thing here but they've never actually worked on a project with people and seen something through from beginning to end. In our situation we had Narbacular Drop [a spiritual predecessor of Portal], which was a clear proof of concept so we knew what we were doing.
Do you have any advice for any students looking to break into the industry?
I think that if there's any word of advice, it’s to make games - as many as you possibly can, because honestly, practice makes perfect. The more practice you get the better off you are when you try and get a job.
I know Valve definitely likes to hire from the mod community if they see someone's made a really interesting project there. There's so many people at the company there, like the Team Fortress guys, and the Day of Defeat guys, CounterStrike... it's definitely worth it to put yourself out there and try and get as much experience as you can.
What was the main thing you learnt at Valve when turning Narbacular Drop into Portal?
The main thing is that playtesting is so incredibly valuable to the process. When we developed Narbacular Drop we playtested a couple of times towards when the game actually had to be handed in, a few weeks before the end of the project, and we discovered all these things that we'd made mistakes on and we weren't able to address them in time because the project was due.
One of the things we learned at Valve is that playtesting is extremely important and that it'll help you craft a better player experience. We started playtesting Portal literally the week we got there; we had the first room of the game essentially, and it looked horrible, but we were already starting to learn how to better our process through those early playtests.
When we were developing the game we tried to playtest every week as much as possible, and when we would design a level or a piece of exposition we would try and get it in the game as quick as we possibly could, within two to five days, so we could see it right away and see whether our decisions with the game were the right ones. It’s kind of subjective until you throw a bunch of playtesters at it and watch them fail and get angry and then you realise 'maybe that wasn't such a good idea, I think we should reconsider that'. It's really good for keeping you objective and concerned about the players.
Portal was in development for almost two and a half years - that's quite a long time for a short-form game. Was that because of all the playtesting?
Actually, I think testing speeded us up because we could quickly determine whether something was a bad decision rather than spending months and months on an idea that was bad in the first place. But to address your question, well, we were students - we came in straight from school, and we had a lot of learning to do, to be perfectly honest.
We had to learn the Source engine, and man it's big, it's got a lot of stuff in there - getting portals to work was hard! We were able to get them up and running fairly quickly but to polish them to the point where we were happy to ship took quite a while, to get that seamless transition and have people not falling out of the universe. The Havok engine really doesn't like you just punching holes in the wall - it's not cool with that! So, it definitely took a lot of work to get that working properly.
In addition to that, the team was pretty small - at the most there was ten people working on it at any given time, but normally it was five to seven people. Valve was working on all the other Orange Box products at the same time, so we had to make do with what we could to make the game good. I think that also possibly conspired to make it longer.
If you'd had more time, would you have wanted to make Portal a much bigger game, or do you feel that its short form part of the charm?
I think that we were never really deliberate about the game length - we just wanted it to evolve naturally. We never set out to say 'we need to make six hours of gameplay' or whatever. The story arc had a good progression that was tuned through playtesting - we definitely knew when players were ready for the next transition, whether they're ready to escape or whether they've hit that high and are ready to confront the antagonist now. Same with the gameplay too - it's all been finely tuned to where players' boredom thresholds are.
Through playtesting we could say that 'OK, this person is definitely not happy, so we need to make them happy'. So I think the length really came out of that and, you know, we're really happy about it. A lot of us don't have a huge amount of time to play games any more - I know that now I've started to work in games I've only got a couple of hours on the weekends or ten minutes here and there to open my DS and play a little Phoenix Wright.
It's nice to make a game that someone can play from start to finish in an afternoon and have a complete experience. We're really happy about it, we definitely don't regret it. We really wanted people to see the end of our game, because we had fun making it and we wanted everyone to hear GlaDOS and hear the song. So, yeah, we definitely didn't want to scare players away from seeing that.
Did you ever think the game would be so popular, and that elements of it would be so memetic?
I think essentially we had a ton of fun making the game and I think that gets reflected in the game itself, people can tell that we had a good time and that we want our players to have a good time too. But no, we never really anticipated the popularity of the game - we knew people were going to like it, but we weren't aware that much. We weren't anticipating people dressing up like companion cubes.