It is very strange to say, but it had more emotional impact than some other things that really try to be epic and sweeping, and narratives that are all about revenge and you have to kill someone because they killed your father.

EW: Yeah. It's always more satisfying for me, personally, when I feel like I can sort of understand the villain in a book or movie and empathize with them a bit. It makes their villainy a lot more tragic.

KS: And human.

EW: And human, yeah. That was the one big rule for writing GLaDOS -- just write her as if she was a person going through a robot "oh my nuts and bolts" sort of thing.

KS: What was nice about the progression was that she got more and more human toward the end of the game.

EW: Yeah, we wanted her to keep getting more and more human until by the end, when the morality core comes off, [voice-over actress] Ellen [McLain] switched to a different voice.

KS: She's our voice actress.

EW: The other thing was we loved Ellen so much and we did a lot of recording with her, so that helped evolve GLaDOS. It was like, "Oh man, I don't want Ellen to say this. Ellen is super-likeable. We should write for that."

It's something you don't always get in games. Because we only have one big character in the game and one big speaking part... a lot of games, you go in with the voice actors, and you have one session with them.

You don't get to do a lot of prep work, so you just do the best you can. But with GLaDOS, over the development cycle, we were actually able to write to the strengths of the actor, which is something you don't always have in the game world.

I would say you usually don't have it.

EW: Yeah.

You may not have been directly attempting to advance narrative and gameplay integration, but I think that's what happened. I would call BioShock and Portal the two games that have pushed it forward by at least trying something that's different. Portal isn't exactly interactive narrative, per se, but it feels like it. Even though the narrative is on rails, the things that you do, she'll still comment on them, like when you knock down the security cameras or things like that. It gives you the feeling like you're affecting it.

EW: We hoped to do that. We had this theory that games tell two stories. There's the "story story" which is the cutscenes and the dialogue, and the "gameplay story" which is the story that's described by the actions you take in the game world. The theory was that the closer you could bring those two stories together, the more satisfying the game would be.

I spent years and years reviewing games, and that's something that always bothered me in games, where the delta between the two stories was real high. I promised myself someday that if I ever got the chance, I'd try to make a game where that delta was almost zero. It was a conscious decision that we wanted to try and keep that world.

KS: It takes you out of the experience, really. You're doing one thing, then all of a sudden the story is telling you, "No, no. You actually did this other thing." "But no, I just did the... all right, fine. You're right, then." I agree with Erik that the closer the gameplay interacts with the story, the more impact it has with players.

Just as an aside, I found it humorous that you're carrying cubes around a lot and it may have had quite a quick "time-to-crate."

EW: Yeah. The crates were in there before I started, so we just tried to make the best of it. It was like, if we were going to have to have crates, they would be the best crates ever.

KS: I guess it would be like a minute and five seconds, in time-to-crate. We trap you in that first room for a minute! (laughs)

EW: It's relatively long. There's an artificial trapping, so we could boost that time-to-crate up. We just call them cubes, though. We never call them crates. It was kind of saddled with it. But then again, it was another constraint, and we tried to make the best of it.

It worked well. You got to destroy the dearest one.

EW: Yeah. It worked out okay in the end.

I had the impression that one of the overarching themes was simplicity, like keeping everything not bare-minimum, but to the essentials, so it never felt superfluous. There was nothing around that really wasn't necessary. Was that something you were going for?

EW: It comes out of playtesting, again. It was the rule we had -- that if enough playtesters couldn't tell us what was going on, it was just too complicated. It wasn't them. It was us. It's our fault that we weren't delivering...

KS: ...entertainment.

EW: Yeah, we're not entertaining. If somebody isn't interested enough to pay attention, then it's definitely not entertaining.

KS: I think there's something to be said for letting players fill in the gaps for what they think they're experiencing, as well. It makes it a lot more fun to have someone dictate to you exactly what the story is going to amount to.

I agree. It gives players credit for having brains, which games often don't do. They're like, "And now you will do this! This thing happened, and here's the backstory." But in this case, I hadn't even really played Half-Life that much, so it still worked for me, without knowing who Aperture Science was.

One thing that I liked, and I don't know if this is intentional, since I didn't listen to all the commentary, was that in the first room, or really early on, you can see the offices. And when I saw that, I was like, "Man, I'd like to get up there." And then at the end, I'm looking out of there.

KS: Yeah, that was totally intentional. Going back to playtest again, we always had comments from players like, "I want to go in that room!" So towards the end of the game when you finally break out, we were like, "All right, we're going to let players see those rooms now." It was definitely fun for us to give that to our players.

Would you say that the design is more designer-driven or playtest-driven? Or is it kind of the same thing?

KS: I think it's one and the same, honestly. It's going back to trying to make the best player experience that you possibly can. At some point, there needs to be inspiration for you to get started and put your ideas out there, but like I said, it's just really, really healthy to stay objective and watch your game be played, because then you'll know if your ideas were good or not.

EW: I totally agree.