Valve on Portal 2

Valve on Portal 2

By Rob Crossley

May 11th 2011 at 9:30AM

Just how did 2011's highest-rated game come together?

This Q&A comes as part of Develop’s package of five interviews with Valve Software. An index of each interview can be found here.

Interview with
Josh Weier (left)
Erik Wolpaw (right)

I’m told that the Portal 2 project was another grass-roots-starter, in that a team got together, with no pressures from management, and began to hatch ideas.

JW:
Yeah it was a really organic process. I mean, coming out of the Orange Box, a few of us still had some ideas of where the game could go. And having finished that massive project with Episode 2, Team Fortress 2 and Portal, it was an ideal time to recharge our batteries and riff a little.

I remember at the very start of the Portal 2 project, there was about four of us, and we all playing with different gameplay mechanics. So we roped Erik into helping out with it too. At its peak we had about sixty people working on it.

EW: It stayed small for a while, yeah. And I imagine you get this now that, at Valve, that projects don’t happen because of some top-down decree, but because there’s enough people interested in doing a project and those people – quite simply – start talking to each other.

I think the reason why there was a lot of love for the idea of a Portal sequel was because just as we were finishing The Orange Box, y’know, a lot of people at the company had been heads-down on one of the three new projects.

Portal was this small thing, by comparison, and a lot of people here at Valve didn’t really get the chance to take a proper look at it until it shipped. So their first-time relationship with the game was the same as the fans’. I think that’s why eventually we got a lot of interest internally about working on the sequel.

At the same time, people who worked on the original Portal, like me, were just really energised by the response to the game.

I’m still trying to understand how games are made at Valve, and because the whole process is so unique, I was hoping we could go over all that in detail. I’d like, essentially, the step-by-step story of how the project was organised.

JW: We had been kicking around game concepts a little bit after Portal and I think Gabe came up to us and said there was a lot of opportunity in making a sequel, and he told us some things to think about in designing a sequel. He was really good at saying what the final project should have, but these were simply things we had to think about and not directions.

So we put together a few concepts that were shown to a few people, just one or two, internally and we got feedback. That’s been the process throughout; make something, get people to evaluate it, edit and carry on.

So we would come up with a new mechanic idea, we’d code it up, we’d draw a map around it and we’d sit people down.

This is a demo you were building?

JW: Yeah, well I guess the best term for it is a ‘test space’ because unlike a demo it doesn’t look pretty, it just has the mechanics. It’s just a puzzle with Portal art assets. We dug into people’s reactions to the game, what they liked about it, what they didn’t like, how they felt about it.

So as we progressed, we started coming up with numerous new play mechanics – finding out what people liked and what they didn’t like. And at the same time, other people sat together and started thinking about the story, how we could move on from the first game. When some of that was put in place, we did some rough voice-over work for Glados to add in the game.

So, with only a few of us working on this, we had built some levels, made some mechanics, created some art assets and had a rough narrative in place. What we then do at that stage is a thing called an ‘Overwatch’.

An Overwatch is where we get quite a few people together for the first time to really test what we have. I think in all we had twelve people look at it. These twelve people are really product-focused – they’ve either led teams in the past, done marketing or they’ve just been really important to game projects.

So with these people, it’s still the same process. We get them to give us feedback on what they loved, as well as feedback on what they hated. But also they’ll offer us ideas as well. It’s not like they’ll suddenly sit down with us and strip away our idea and add in theirs, it’s more like they’ll say we should look into something.

It’s more a barometer-like process than anything. It’s there so other people can check-in to see what a team is doing, and seeing if that matches what’s good for the company and what’s good for the fans. But also an Overwatch can give teams really useful advice on how to evolve the project.

But it’s a key philosophy at Valve that decisions on a project ultimately have to be made from people working on the game. An Overwatch team doesn’t come back and list all the things we need to change, they simply say ‘this is what we think’ and we do what we want with that feedback.

And of course usually the feedback you’re getting makes a lot of sense and is coming from people you respect. However, and honestly, we can also say we don’t agree with certain points and the company will understand that.

EW: In the case of Portal 2’s first Overwatch, some designers who loved what we were doing jumped on the project a bit later on.

So people can say, ‘I loved it, can I help you guys make it’?

EW: Yeah completely, and in fact that happened a lot after Left 4 Dead 2, in that people finished that project and jumped onto ours. Often people will stick to their own project, unless they’re a free agent, but usually people here are attached to something.

How does that work? When people finished Left 4 Dead 2, how far into Portal 2 were you?

EW: I think we were about 2 years into the project. But I should say, when Left 4 Dead 2 was at its final stages, and in fact when the first Left 4 Dead was, a bunch of Portal 2 developers jumped on to those projects for a couple of months to help finish them up, and then essentially came back with extra people in tow.

How does that work? How do people jump projects half way through?

EW: It’s not too hard, I mean there’s still a lot of input to make and there are still creative decisions to be had – so it’s not like you’re jumping across to execute on some plan. Y’know, stuff changes on Valve projects all the way up until the end, until you reach a stage called ‘Content Lock’, where people aren’t making content changes but just fixing bugs.

But still, before a Content Lock there are lots of things you can change and lots of improvements you can make to a game. I mean, we were granted an eight week delay on Portal 2.

I know it sounds so obvious, but the people working at the end of a project are so much better at their tasks than they were at the start. People kind of optimise – eight weeks at the end of a project are ten times more useful than eight weeks at the beginning. So, everyone’s firing on all cylinders by the end of a project, so if we can get a few more weeks, I mean there’s a ton of things we can change and improve.

JW: It’s kind of a weird thing, because there isn’t a set pattern of making games at Valve. There is of course a cabal and an overwatch to help us out, but every team and every project comes to solve every problem differently. Portal 2 is a completely different game to Half-Life 2 was, and if we tried to replicate that process it simply wouldn’t work.

It’s just so different from other studios. You just said ‘if we can delay it’ like it was a casual option.

EW: Well when you’re not beholden to a publisher, there are less reasons not to delay a project. Of course there’s some marketing considerations to think about. If you start buying ad space and things like that then of course it becomes tougher.

But by the time we thought we should delay Portal 2, Valve hadn’t been booking ads for it yet, so we went up to marketing and said we need eight more weeks. We asked them if April was any worse than February for any particular reason, and at the end of the day no one had any significant reason to not say let’s just go with April.

JW: I do remember meeting up with Gabe at one point for a bit of help and advice, because, y’know, this was one of the first big projects that I had to help manage. And he said to me, ‘I could take hold of the handlebars and help you steer but I would probably knock you off your bike much more’. So his approach, and the company’s at large, is to help when help is needed but to let the project drive itself.

Was Gabe in the loop at any point during discussions to delay?

JW: Yeah he was in that meeting. He was actually the biggest supporter for the delay. He was asking about what we wanted to do with the time, and he said there was no data to suggest a delay was going to hurt us, and that in fact all the stuff we said we wanted to do for the game made it sound like it was going to be really good.

EW: It was not especially controversial, yeah.

JW: And it wasn’t like we were racing to get the game out the door before the fiscal year-end to please our stockholders. Our fans are our stock holders, really. What really would have been a controversy would have been to get a game out early and it not being very fun.

EW: In fact, I think the only negative approach to the whole delay thing was people asking if the team could handle another eight weeks. That was carefully examined because, of course, you do hit a bit of crunch time there.

If this game hits a certain Metacritic score, will you be rewarded?

EW: Well, because there really isn’t a hierarchy at Valve, there’s a lot of value in having made the right decisions on a project, in terms of how other people view you.

Obviously it’s a business and we get bonuses and the occasional raise. But no one here that I know of has a contract that says ‘get a Metacritic of 92 and you’ll get this bonus’.

JW: Damn, that would have been a good idea though.

EW: Yeah [Laughs]. But every year there’s a peer review here as well, and it keeps everyone hard-working and honest about things, because it’s especially important in a place where there’s no hierarchy that you still get feedback on how you’re doing.

It feels like we’re not explaining it correctly. I mean I’ve been working here for six and a half years and I still can’t explain to my wife how it all works. She thinks it’s the craziest system for making things. She tells me it can’t possibly work, and yet it does, and in fact it does better than any system out there.

It’s almost one of those things I don’t want to put my finger on and explain too much, because it’s ephemeral and if we explain it, it will go away.

I suppose a big part about all of this is the hiring process is relatively gruelling, because Valve tries to find the best self-directed people. Pretty much everyone on a single project is equally as responsible for it being good or bad.

No one feels like they’re being told to make shitty textures, and no one here is doing their work resentfully thinking ‘one day I’ll finally be able to make my own game’. Everyone owns their project, everyone carries their own responsibility.

It’s almost a Marxist approach, in truth.

EW: Yeah, I hate… yeah. Just please put this into your interview, it is quite Marxist, and it does work, but no, I’m not saying it would work governmentally on a larger scale, okay? No. In a tiny environment like this it probably works, but not in this country, okay, and it’s probably is on the verge of collapsing anyway, okay, let’s get that out the way. I don’t want some bomb-throwing anti-Marxist on my case. [Laughs]

If someone steps out of line, if someone takes on too much of a project, or becomes too demanding, or is simply a bit of a jerk, what do you do?

EW: There’s almost self-correction built into the system. Making a game is such a complex thing, that there’s only so much you can do on your own, and so if you go off the rails too quickly, there’s not really going to be anything to do.

And people are not going to do what you tell them to do. You either convince them to do it or die on the vines.

JW: I think there’s a lot of collaboration built into our projects, so usually that doesn’t happen at all. That only happens really if someone goes away into a corner and starts building a load of stuff without talking about it.

If there’s a design problem, we’ll pull all people in to discuss it – not just designers. I mean, you need the feedback from the artists, right, otherwise your solution may just be a problem carried over to them.

So we pull together, talk about it, and everyone has insight on where we are. And everyone here is encouraged to talk about and show what they’ve been doing. We push it all the time, get feedback as soon as possible, show people what you’re doing. Like I said, people get in trouble if they sit in a corner for too long and do a bunch of stuff without telling other people what they’re doing.

I think you learn here very quickly that you need as many eyes on your own work as possible.

EW: I think we develop a relatively thick skin for accepting negative feedback. That’s just part of the system. Because the hiring process is so strict, everyone here all know that everyone else here has passed a bar. The person giving me feedback, even if I don’t know much about them, is probably very talented.

JW: At the company I worked at before, everyone was trying to climb the ranks and get as high as they could. But we have people here, like project leads, who really don’t want to do that at all. They just want to make some amazing things and distribute that out. But that’s the funny thing, you could pick five people in a team would could step up and lead the group straight away. So that helps a lot too, because if one person in a team gets over-burdened another can just step in and lead things for a while.

EW: The last place I worked was completely different to this, especially in regards to how I was one of the eldest people at my last studio and that’s not at all the situation here. There’s really no burnout.

Valve has found a way of making games a job you can have with enough time to have a family. Of course, we have to work hard, but it’s not the endless grind. I think a big part of that is how successful Valve is commercially. I think if money was tighter there would be more constraints.

It’s interesting you say there are a lot of older people at Valve, because the industry standard is to exploit young people to the point where they mentally and physically have enough of it.

EW: Yeah and they go off to work for some database software company somewhere. Once you get married, once you have kids, in your head you realise it’s just not worth it. It’s just not worth it. The fourteen-hour days, the stress, it’s just not worth it.

Did life get easier when you joined Valve?

EW: Yeah absolutely. I worked at Doublefine a while ago and I think even they would admit things got a bit crazy. I was working on this project called Psychonauts and it was horrible.

It was my first experience working in the industry, and I was done with it. I quit the company and I was pretty much ready to quit the industry – it wasn’t fun, it was interesting, but I had no real desire to do that again.

Then when I got a job offer from Valve, honestly it was – I was doing some contract work here for a month, and this was a year after I quit Doublefine – and my wife was saying to me, ‘really, you want to go through all of this again? You want to move for another game job?’

I must admit, I thought, well, it could be just as bad. But it was night and day. It was nothing like before.

I know Doublefine got way better, obviously.

JW: The place I worked before, they essentially hired my when I just finished High School. That was their thing, they would hire young talent and just burn through it. The churnover was ridiculous. It’s interesting because at Valve only rarely do they hire people with little experience.

EW: And a weird thing is how many people working on the first Half-Life are still here. There’s way more of that team working here than there isn’t. I mean, John Guthrie, he was working with us on Portal 2.

The hiring process must be ridiculous.

JW: It’s a whole day thing. What is it, six interviews?

EW: Six interviews, of people in different disciplines.

JW:
I actually failed the first time. I was interviewed when they were looking for people for the Half-Life 2 project – they wanted people with good FPS experience. And I had worked in this field a lot, and I remember I sent Gabe an email after the original Half-Life because I was working on the Heretic 2 and Half-Life just completely squashed our game. We were kind of mad but I was amazed by Half-Life and I just wanted to say I was impressed.

One day I get an email asking me if I wanted an interview. I was a kid at the time and to be honest, it was brutal. They throw challenges at you that are out of your speciality.

So anyway, I had failed spectacularly, I licked my wounds, made some more games, and asked for a second interview.

EW: That is the thing. I’ve seen it a lot. When people are not hired from interviews, that doesn’t mean they’re out the equation, it just usually means they’re not ready yet. But I’ve seen it happen a lot here – people not getting past the interview stage, coming back, and making it through.