I was at Valve last month to interview pretty much everyone I could find, and play one of the most exciting PC games on the horizon: Portal 2. The preview I wrote, and the profile on Valve themselves, is in the new issue of PC Gamer in the UK . But we're also putting up the interviews here on the site , one a day for a week. Today's is my conversation with Gabe Newell, Erik Johnson and Doug Lombardi on how they test their games, how they can see your pulse race, measure your stress and sense your sweat. And how they'd like to use that in a game.
PC Gamer: Josh [Weier, project lead on Portal 2] was telling me you guys had one of your accountants play Portal 2 just to see how she got on with it, not being a gamer. Is that an important part of the testing process?
Gabe Newell: Yeah. With any game you want to find people who've never played games and put them down and see what happens. You learn a lot more from those people. You need fresh eyes, always. Because we live so much in this space, it's interesting to be reminded that some people don't know what a rocket launcher is. Like, literally don't know what they're supposed to do.
Erik Johnson: Having someone that never even intends to play or buy a video game is useful because you've still got them to make your game so that they're entertained if they're that far out of it that you will definitely learn a bunch of things that are really applicable to everyone else.
Gabe Newell: Yeah, and the lessons you learn are applicable much more widely than to the specific demographic they represent.
Erik Johnson: Yeah, there's this common misconception that if you take this unskilled or junior video-game player, that the result of that will mean that you change the game to make it easier for unskilled players. It always benefits all players.
Gabe Newell: It forces you to have a lot of clarity. Not change a skill curve or something.
PC Gamer: I had my dad play Portal 1 once, and he completely grasped the portal thing, but he couldn't get the hang of changing his angle of view and moving at the same time. So he'd walk along, then use his mouse to look around, walk along and so on...
Gabe Newell: I remember back when nobody used the mouse to play, and that was this huge scandalous thing. And Romero would get on and say, “I'm not going to use a mouse.” (Laughs)
Erik Johnson: We had a playtester from Half-Life 1 who – I don't even know his name, he ended up being in the games industry - but we nicknamed him Keypad Boy, because he used the keypad to navigate everything.
PC Gamer: I must've used it, like... two years after it was cool. Because in the Build-engine games, you'd look up and down with the mouse and the whole thing goes [cry of existential terror at the world distorting], because they had no vertical vanishing point.
Doug Lombardi: Yeah, you could tell the Duke players from the Quake players...
Erik Johnson: They'd all have inverted mice.
PC Gamer: Yep, I invert my mouse.
Gabe Newell: Well, Robin for a long time was an inverted mouse guy.
PC Gamer: I very rarely hear of people switching actually. I've tried to switch...
Erik Johnson: If you work at this company, and you're hazed for long enough, you find you switch. (Laughs)
Gabe Newell: Have you met with Robin (Walker, project lead on Team Fortress 2) yet?
PC Gamer: Not on this trip, no. (When I did, he told me about the future of Team Fortress 2 .)
Gabe Newell: Remind him he's Australian.
PC Gamer: Maybe that's why, actually.
Gabe Newell: You should make that joke, say “I heard that you use reverse mouse. That makes sense because you're from the southern hemisphere.”
PC Gamer: (Laughs) So it's come to that kind of discrimination?
Erik Johnson: It's not discrimination, I mean technically you're his superior right?
(All laugh)
Erik Johnson: Just ask him “Whose Queen's on your money?”
Gabe Newell: Ask “Where is the silverware anyhow? We're still looking!”
PC Gamer: Gabe, I think last time we met you were...
Gabe Newell: He's going to try and turn this into an interview.
(All laugh)
PC Gamer: I have to! You described yourself as a cheerleader and chief playtester. What do you actually find yourself doing these days?
Gabe Newell: These days? So my job is always changing, right? That's the nature of the industry; it changes for a lot of people here. So right now I'm thinking a lot about longer term stuff, I'm thinking about thin client architectures, I'm thinking about cross media authoring, I'm thinking about the problems that game companies have getting movies made out of their games, and movie companies have to get decent games made. What else am I thinking about? I'm in thinking mode... oh, I'm also thinking about biometrics. Sorry, that was one thing I forgot.
PC Gamer: What does that mean?
Gabe Newell: So when you look at our games, more and more we have this representation of player state, where we think we know how you feel, essentially. And with biometrics, rather than guessing, we can actually just use a variety of things like gaze tracking, skin galvanic response , pulse rate, and so on.
Through combining those pieces of information, we can get a much more accurate indication of player state. So that's something we're super interested in. We've done some experiments in that space, and feel like there's some easy wins for customers and for developers.
And then there's some surprising side-effects that we didn't expect, like what happens when you expose that information in a social gaming context. It surprises us that how much value there is to the people who are playing. So if you're in a competitive situation, and you see somebody's heart rate go up, it's way more rewarding than we would have thought. And if you see somebody in a co-op game who's sweating, people tend to respond to that way more than we would have thought.
So we can stop using our guess at what your player state is in Left 4 Dead, that we kind of expected. But the value of being able to see what other people's biological state is in social gaming, that was not something we were anticipating. But that's just the way things go.
PC Gamer: This is for internal testing right? You're not going to sell me a heart rate monitor and plug it into Steam?
Gabe Newell: Well, what you want to do is figure out how clients can expose their state. So you're trying to find non-clunky, non-stupid ways of getting that data. I mean, if you sit there and give a medical technician 30 minutes to wire you up, you can get awesome, awesome data. But it's just not the consumer experience.
Doug Lombardi: We can ask them to shave their heads before they play...
PC Gamer: That's a small price to pay.
Gabe Newell: Right. But if someone comes up with a clever way to take some non-visible light and bounce it off your retina, and read it with your web camera, and get your pulse rate that way, then that's pretty cool. Because it may be a hard problem, but if you solve it once then you're done. It's not like a recurring hard problem.
So we think there are several people out there with interesting approaches on the hardware side. Enough that we have confidence that the hardware side will be a sort of resolved problem in the not too distant future. So we need to figure out how to take advantage of it.
PC Gamer: Right now, how much do you track of your tester's behaviour from a computational side? Do you track exactly where they look with their mouse at all times?
Doug Lombardi: We're getting gaze tracking now.
Gabe Newell: You mean internally?
PC Gamer: Yeah.
Gabe Newell: Oh internally yeah, we do gaze tracking. We've got this $50,000 system – it's a lab system for doing gaze tracking.
Doug Lombardi: That was way more valuable than we expected it to be, influencing work pretty much immediately.
Gabe Newell: So when you're designing a game, you think every pixel is just as important as every other pixel, and you certainly expend effort that way. Which, when you think of Half-Life 2, is sort of ludicrous right? You're whizzing on this boat, and you're going down this river, and then something's falling in front of you- and somebody worked on the texture on the door over there and you spent one billionth of a second even being aware of that.
With gaze tracking it's even worse than we could have ever imagined. A huge percentage of the stuff we draw on the screen people never even look at. And so what you want to do is use that and redesign it.
Your first reaction is, “Oh man, we're not designing these things right, because if they're spending all their time looking this rectangle on the lower half of the screen. Maybe that should just be the screen?” So you want to actually provide meaningful stuff on the screen. And even then, if you end up finding that people spend most of their time looking here or here, then obviously you want to allocate your rendering quality or whatever 'budget' you have that way.
So I think we'll move from the era of homogeneous allocation of screen real estate to rendering performance and visual quality, to a much more accurate [system where] the things that you actually look at are the things that'll be drawn the best.
People know that when they play games, they recognise that the little dragon that's on the edges of the screen becomes invisible after a certain point in time. So nobody actually spends any time drawing those things. But there are things we spend a lot of time drawing that just don't matter to the player, so this just teaches us that that's waste.
PC Gamer: I get a degree of players guilt when, in Half-Life 2: Episode Two, I'm driving past all that gorgeous scenery. I'm thinking, “Oh this is gorgeous, it must've taken someone ages to design each little metre of this,” and just blowing through it.
Gabe Newell: Well, I think that's totally a design flaw, right? If it doesn't matter, people don't look at it. So the challenge is to create stuff that has significance. If it doesn't represent a choice for you, then why waste time building it? Games are about phenomenological difference, not about scenery. So unless you're putting logical information or choices into that stuff, then you're making a mistake.
PC Gamer: Episode Two sprung to mind, but that's actually only a brief section where you're driving and there's nothing in the scenery. Because once you get the scanner in your car, that scenery is presenting you with a choice. You can stop and explore it looking for the caches of goodies.
Gabe Newell: A design experiment I've always wanted to run is: nothing can be different unless it matters. So everything has to be exactly the same colour unless there's information there. And just see where you end up. I think that'd look really weird at first, it would be like the first time someone saw abstract expressionism. “That doesn't look right.” But then after you've played it for a while, you suddenly discover that everything that was different had significance.
PC Gamer: There's interesting little hints of that in Portal 2 - and you did this to an extent in Portal 1 too. Sometimes there's just a mark on the wall that, in terms of the fiction of the game, is decorative or just an accident. But in terms of the game logic, that's a good place to put a portal. You don't have to, but if you're already thinking about putting one around there, you instinctively focus in on the exact right place.
Gabe Newell: I think we're getting better about that, actually. It's just for us in our copious spare time to do the experiment of: “OK, you can't spend any time on anything unless [it has significance]," as a hard rule, and see where we end up.
In our final interview tomorrow, I ask Gabe what the piracy rates are like on Steam games, and what's going on with Episode Three. You might not like the answer.
