And with the tagging and liquid stuff, what would you say was the decision behind having that be an environmental effect versus something that you actually shoot out in the original [student game] Tag: The Power of Paint?

EJ: Most of it was just playtesting. We had lots of different approaches on how to change the state of a surface, and this was the one that ended up fitting the best.

Did you actually prototype different kinds of implementations to see if they would work?

EJ: Yeah, the player shooting it -- yeah, lots of different types of surfaces.

To me, it seems that having to shoot anything other than portals would probably add too much complexity.

EJ: There are a lot of advantages of having a game where the gun has the number of states that the portal gun has, especially in terms of players that don't play a huge number of games. It is a great interface with the world that always acts the same way. 

The physics of the original game could sometimes feel complex, and it's sometimes easy to forget which portal is which even though it's so simple. Are you concerned about players keeping it all straight now that the title will be longer and incorporate more elements?

EJ: It's not uncommon for people to use the wrong portal when they are doing it, but take the things that look like tractor beams [in <i>Portal 2</i>]; there's not a huge ramification for the player doing that, because from their vantage point they can generally replace portals. The worst that happens is that they have to re-do it. The portal gun does light up to indicate the last portal you put in the world so you can try to keep track of it, but Portal definitely bends playtesters' brains a little bit.

The trailer I was watching was showing more advanced gameplay, but would it be correct to say there are more timing-while-moving based puzzles? It looks like you have to bounce here, you have to grab this thing while it's in the air and you have to get over there.

EJ: I don't think that would be a true statement actually. It would be a true statement to say those types of demos are interesting to watch to people who have never experienced Portal. It's like the idealized ninja -- we call them "ninja moves" internally.

The Portal experience Portal fans have, where it's surprising in this interesting way -- everyone who liked Portal 1 had that, but it makes for very bad demos at trade shows, since it's kind of slow. It's challenging in the same way, and perhaps even more so than Half-Life 2, where we are talking about the stories, so it's near impossible before releasing the game.

Talking of story, GLaDOS is back, and the comments have been revealed for her this week make her sound like a jealous ex-girlfriend. Is that the intention?

EJ: Yeah. You are the only person she can have interaction with, but the problem is her only way of interacting with anyone is to test them. She can't really kill them, but she can test them. She has no point of being around if you are not around.

It's an interesting dynamic because there's this love-hate relationship going on there. I know the player is just an avatar, really, but what have you discussed internally about what the player feels about GLaDOS?

EJ: Well, there's what the character Chell thinks about GLaDOS, and what fans think about GLaDOS.

I mean the in-game character, yes.

EJ: We haven't really explored a huge amount of what her relationship is. Generally, it's more that the customer would think of how they think. We did a pretty good job of getting customers interested, which was our goal, and GLaDOS is generally not hated in the way bosses are typically hated in video games.

Even bosses that are executed really well, most of the time their goal is to feel like they are a direct antagonist and your goal is just to defeat them. GLaDOS definitely doesn't have that.

Is the aim, if such could be identified, of Portal to escape? Is it the same? Does it matter if it's the same? Having two games with the same end-goal -- I mean, maybe that's not the end-goal, but does that matter?

EJ: Yeah, I think it does matter, but I think the implementation matters a lot more. This is getting tricky to talk about, because it's about story stuff a little bit, but I think if you are telling players that the core of the story is "you are going do again what you did last time," for most people that is pretty unappealing. That's not what is going to happen in the game, but there are definitely some things that are similar to the previous game. In implementation, they end up being fun and different. You're still going to have a testing relationship with GLaDOS.