Half-Life: Alyx, Valve Software’s flagship virtual reality game and a long-awaited return to the Half-Life universe, was released recently to near-universal acclaim (you can read our review here) and high streaming numbers.
Alyx was developed in the Source 2 engine at Valve Software, based in Bellevue, Wash., over the course of the last four years. It’s the first project to materialize out of Valve’s stated plans to make three first-party, high-budget “flagship” games for virtual reality.
Thanks to COVID-19, a planned press event for Alyx in Bellevue on March 20 was canceled. Instead, I had the chance to discuss Alyx’s development with Valve’s Robin Walker and Jane Ng via Skype.
Walker is a 22-year veteran at Valve, starting with his work on the original Team Fortress mod for QuakeWorld. He’s been a programmer on several of Valve’s highest-profile titles like Half-Life 2 and Left 4 Dead, and was the co-lead designer on Team Fortress 2.
Ng has been in the video game industry for almost twenty years, working as an environmental artist on games like Costume Quest and Brutal Legend. In 2016, she joined the newly-founded developer Campo Santo, where she designed the environment and produced the hit indie game Firewatch. Campo Santo was subsequently purchased by Valve in 2018 and relocated to Washington state.
The following interview has been edited for clarity.
GeekWire: So you’ve been working on this full-steam since 2016?
Robin Walker: We started in February of 2016, I think, with a small team, and we brought out a small prototype. Then people started to play that, understood what we were trying to do afterward, and started joining up.
We had 80 people on the team when we were about midway through. The exact size of the team I wouldn’t be able to tell you. The way things work at Valve, people organically join once they’ve finished up what they were doing before, and if what you’re doing makes sense to them.
So it was always full steam ahead, I guess, but not in the sense that all 80 people were there from day one.
Jane Ng: I joined the project last year, I think. People just sort of see that “Hey, this project’s getting pretty cool,” and then they roll their desks over when they’re done with whatever they were doing.
GW: You’re on the Campo Santo team. Is In the Valley of the Gods still happening?
JN: It’s on hold. We jumped onto Alyx first. That’s kind of all hands on deck right now.
RW: I think one of the things that’s confusing for fans, and people in general about Valve, is our unwillingness to ever really say whether something is completely on or off the table. It always sort of bums me out, because it’s not an attempt on our part to be obtuse. It’s just the accurate representation of the way we work. We really don’t know what we’re going to be doing a year from now.
I think anyone who thinks they know exactly what the state of the video game industry is going to be like five years from now is quite likely to be wrong. One of our strengths as a company is that we don’t have any external investors. We’re fully self-owned. We have the freedom to look at what’s coming, look at what’s changing, and try and adapt to that. We don’t like to make plans that go too far out because we couldn’t maintain that flexibility.
If you go back far enough, you can find us saying how we weren’t sure we were going to make single-player games again. That’s a laughable thing to go and read, given what we’re shipping, but we don’t regard that as a failure on our part. Statements we make at any point in time are an accurate reflection of how we feel at that time. Us hedging is just us avoiding saying something that we know is just not actually true.
GW: Well, that’s the downside of having created some of the most passionately admired IPs out there. In anyone else’s hands, Half-Life would be on its ninth sequel, a second Hollywood film, a breakfast cereal, a cartoon show with adorable headcrabs…
JN: I want to see that.
GW: I guess there is that debeaked one in Half-Life 2, isn’t there.
RW: Yeah. LeMarr.
GW: Anyway, it’s been 13 years since the last Half-Life game, which feels like an eternity in Internet time. I was curious how intimidating it was to try to get back into developing Half-Life at all.
RW: “Intimidating” is one word, I guess. I think that speaks to what you were saying about how in some other company, we’d be on our ninth sequel.
Half-Life products have always started with some opportunity or some technology that we felt were really interesting, in terms of the way we could use it to create some new experience for the player. If you go back to the first Half-Life, we were looking at first-person shooters at the time. They were fun to play, but very focused on combat. We thought there was a real opportunity in that medium for being a way to deliver stories that perhaps hadn’t really been done before. Less perhaps a focus on the story itself, but the way that you can tell stories in that medium.
Then you go forward to Half-Life 2, and there were a couple of opportunities there, as well. We really thought that there was an opportunity around the way we were representing characters, people, in video games and in those sorts of narrative approaches that we’d started in the first game. So we worked on the technology there, and at the same time, it was the birth of the sort of physics engines that eventually went on to become things like Havok. We thought there were real opportunities for gameplay with those kinds of technologies.
I think one of the problems over the years after Half-Life 2’s episodes is that we were always searching for that next opportunity. It’s a hard thing to work on a product – you want to think about the word “intimidating” – it’s really intimidating to work on something that people want to be Half-Life 3, but you’re just searching for that opportunity.
It’s hard to start writing code on that today. It’s hard to create art on that today. You have to accept the very real possibility that you do that for six months or a year and at the end, still have nothing. That’s really scary.
Whereas on this project, I think we cheated on that. With Half-Life or Half-Life 2, we started with the opportunity at the forefront. We knew what it was right away. We’d made a prototype of Half-Life 2 in VR, and realized that a lot of Half-Life 2’s core gameplay experience is multiplied by the benefits you get from moving into VR.
The challenge became, “take Half-Life 2 and make it work in VR.” That’s something I can write code on today. There’s no real risk that six months from now, we won’t have something that’s six months cooler than we have right now.
I think that was the intimidating factor that made working on Half-Life and integrating stuff terrifying. That was just not even there on this product at all. I really just started working on something right away, and once you start working on something, you can start putting it front of players and seeing how they react. Once you’re in that phase, it’s just sort of a matter of time, really.
GW: That feels like it goes back to the freedom of Valve that you were talking about. You don’t have to worry about, for example, your shareholders.
RW: Yeah, exactly. You can just focus on bringing people in who are proxies for players, and see what they do.
When you think of something like Counter-Strike or DOTA 2 or some large service game that has a large multiplayer audience, and you’re going to ship something every week like those teams do, you’re focused on the larger systemic designs. You’re building stuff and getting it out there to that larger sort of audience and looking at the data coming back.
With single-player stuff like Half-Life, we’ll go into a room and design the next space that the player enters, and what happens in that space. We’ll talk about what we need to do, and then we’ll break apart, build it, put it in front of a player, and see what happens. You iterate on it, and then you go make the next room. You just keep doing that until you’ve built the whole game. It’s very entwined in the regular playtesting process.
I find it fun to work that way because you get to be so responsive to what players do. The second time a playtester tries to do something, you say, “We should make that work.” Huge parts of your game just creep around the design as a function of what you see players doing.
JN: Especially in VR, too. What you think works isn’t really what ends up working, a lot of the time. At least for the artists, we have to kind of relearn how to make art just for VR, for it to look the way you think it should look. You see it in VR, and it’s like… “Oh, actually, it doesn’t really quite work.” We had to figure out how to make it work in VR instead of just on a flat screen.
GW: Because VR’s much more manipulable?
JN: There’s that. Just for example, I think in VR, the scale of everything matters a lot more than you would think. Making a space feel real is much more important than other things, like just how pretty everything is. The kind of art that creates the kind of presence you want in VR is very different.
RW: Yeah. One of the first things you realize when you start to put players into your VR game is that they’re all different heights. We play-tested on younger kids, and we have people on the team who are six and a half feet tall. You’re watching the screen that’s representing what they’re seeing, and you start to see the world through the eyes of people with that sort of height difference. Our world has to support that; that’s a real, significant thing.
The first thing everyone says as a reaction to that is, well, why don’t you just make everyone the same size? Which we could do adequately very easily, but it’s a terrible experience as a player. You need to reach down and pick things up, for one thing.
You’ve spent your whole life learning how big a table is as relative to you. If you suddenly enter a world where the scale of things doesn’t relate to your real-world experience, it’s all just wrong. It doesn’t work.
GW: I’m thinking about how motion-sick I would get if I was suddenly a foot taller, or a foot shorter.
RW: [laughter] It results in all these weird things, too, where we’d have players say things like, “Combine soldiers are shorter than I imagined.” It’s like, “You’re six feet tall, dude!”
I remember one of the more entertaining bugs was one where one of the players on the team was complaining. “Is anyone else having the problem where they just keep dying for no reason at all? Like I’m just standing there and I die.” Everyone else was saying that they didn’t have any problems.
We dug into it, and realized that he was so tall that the distance check that the game was doing, to see if players were falling to their death, was triggering all the time. His perspective was so far off the ground that the game basically thought he was falling all the time and he’d die from impact. So many little things.
GW: How tall was he?
RW: I think he was only six and a half feet. Those sorts of things get tuned around the height of the programmer who tuned them, who wrote the code.
GW: That’s a really interesting detail.
RW: It’s cool because you’ll watch a kid play, and they’ll find all sorts of things in the game more easily than a taller person. They’ve got an angle that goes a little under so many of the things in the environment.
Then you’ll see the opposite, from taller players. They’ll see a piece of loot on the top of a shelf somewhere that we’ve positioned so it’s just peeking over the edge for those of us who are average-sized. But for them, it’s right in front. It’s been an interesting thing.
JN: In traditional games, I think a lot of artists build to the camera. You know exactly how much screen the player can see, and then you sort of pack everything you want them to see into where most people would see it?
But in VR, you can’t really do that, because your head is the camera. People do look under things, and go behind things to look around. The way we populate the levels changed quite a bit. I think it just makes the game more fun that you can find the little things that people hide. It’s very different, working on a VR game.
RW: All the features of the game, what you do in it, is really a function of things we initially thought would be fun based on our experiences. When we first built that prototype, we started to explore what we were enjoying, in the differences between Half-Life 2 on a flat screen vs. VR. Then, as we start to build the track, and players start to play it, we started looking at what they were doing and responding to it.
Early on, it seemed pretty obvious that exploring the world was more interesting in VR than it had been before. We intended to lean into that in terms of how you would find resources and stuff like that, but as we started to playtest, players would do that to a greater extent than we thought they would. So we started putting more of that sort of stuff in, and then they would do it even more.
The level and density of art in the world just kept increasing. One of the first combat areas of the game is a place we built three-and-a-half years ago. It’s been play-tested to death, and it went through, I think, almost three discrete art replacements. We got a better and better sense of the degree to which people wanted to explore the world, and so, every time we put more detail in, they paid more attention. So we went in and put in more detailing.
JN: Opening lockers is so fun in VR, and you never would have guessed that. We love opening filing cabinets, and all the other things you can open.
GW: Valve told me that Alyx is about 15 hours long?
RW: Yeah, it takes people somewhere between 15 and 18 hours. It depends on how much attention you pay and how much exploration you do.
GW: That’s one of the longest VR single-player experiences I’ve heard of yet.
RW: Yeah. When we started, that was one of our core goals. There were a bunch of people working in VR around the time when we started, a bunch of devs doing all sorts of interesting and creative stuff with mechanics, and the community was having a conversation about what they want to see in VR games.
It was really clear that a bunch of people agreed they wanted a really big, triple-A kind of VR title. There were all kinds of reasons why that would be really hard for a smaller dev studio.
We felt like that was a kind of role we could play. We could justify it to ourselves, and have the freedom to be able to make that kind of choice, to invest significantly in a longer-term approach to building this platform out. People were waiting for this.
Then, on the other side of it, we’re big believers that new technologies on the input side of VR, like tracking controllers, are really empowering from a game design perspective. Tracking control was one of the most exciting new input paradigms that’s come along in years in game development.
GW: I was curious about the architectural and other inspirations you were working from when you were building the environments. City 17 in general always reminded me of sort of Cold War-era city designs; everything’s very stark.
JN: Our art team were definitely using a sort of Eastern European basis for reference. In Alyx, though, there are actually a lot more different kinds of locales that let the artists be creative in different ways. I don’t want to spoil too much of what’s in the game, but there’s a lot of stuff that isn’t just architectural. You go through quite a few different and unique settings that aren’t just hallways and buildings.
RW: To speak somewhat to the narrative side, this product was interesting because it sits between Half-Life and Half-Life 2 narratively. At the same time, we’re cognizant of the fact that we have a lot of customers who may have played the game 13 years ago. We’ve also got a lot of new customers coming in.
When we built that first prototype using Half-Life 2 assets, one of the most common things we would hear from players is that it felt really cool to be in a place that they remembered really well, but had never stood in it to the extent that VR let them.
We felt like it was a real challenge on this product to find the right amount of nostalgia and novelty, and not to overuse that. We did definitely make the choice to start in a more familiar space, in a place that people are expecting, in City 17, to target that nostalgia. But then, start to move away from it as you play through the game and get to those areas with more novelty. Because you know, you don’t want to overplay nostalgia. You still fundamentally have to see some new stuff.
We used previous games, like Half-Life 2, as a reference to some extent, because Alyx is narratively between these two points. There was an interesting conversation we had, where if the Citadel is like a giant clock, and you can see its end state in Half-Life 2, how does it get there? What are the various other things that are coming along in the narrative at this point? Wallace Breen is not yet at the height of power he has in Half-Life 2; he’s still having to actually fight with the factions to become the spokesman for humanity. There are lots of interesting ways to think about the transition.
One of the things that really helped us there was having a bunch of people who did work on previous Half-Life titles, but also a bunch of people who had come through and who didn’t work on it. We were able to get that mix of fresh eyes and fresh perspectives. In some cases, we had people who were fans of the original, who could draw from those older titles without putting them on a pedestal, where the decisions in it that had to be inviolate and had to be adhered to. Instead, we could treat it like any other piece of source material.
GW: One last thing: to whoever was involved with the decision to not make Alyx a silent protagonist, I appreciate that. Thank you.
RW: I think everyone always assumes that we’re religious about that due to Half-Life and Half-Life 2, but we’re not. We try and avoid being religious about everything; we try to be as pragmatic as we possibly can be.
On this product, we spent a bunch of time debating about whether or not Alyx should talk, and in the end, we are very happy with the decision. It was informed by various things like making sure it was different to play as Alyx than it was to play as Gordon. You’re playing a protagonist who you know and understand; she’s not a blank slate. She shouldn’t attempt to hide away so she can just be you, the way Gordon does.