Skip to main content

We talk Half-Life: Alyx with Valve in "the meanest interview ever"

What it means for Half-Life 3 and where VR goes from here

It's taken 13 years, but we finally have a new Half-Life game. It may not be the long-awaited Half-Life 3, but as you've probably seen from our Half-Life: Alyx review, Valve's first foray into virtual reality shows they're a developer that are still very much at the top of their game. But Half-Life: Alyx isn't just the work of a talented team of developers. It's a game that's ultimately been shaped by the people who have played it - the hundreds, if not thousands of playtesters who helped Valve turn their most famous FPS game into a VR sensation.

To find out more about how Half-Life: Alyx came into being, I sat down with Valve's Robin Walker and Jim Hughes a week before the game's big release day. We talk about everything from revisiting the Half-Life series and the challenges of bringing it to virtual reality, to what this means for Half-Life 3 as well as just whose idea was it to have headcrabs jumping directly at your face. Some of the answers you'll have seen appear on the site over the past week, such as how playtesters became obsessed with collecting every last thing in sight to everyone assuming they were playing as Gordon Freeman until Alyx was finally given a voice, but there's plenty more to discover here as we lay out our chat with Robin and Jim in full. Enjoy.

RPS: What was it like coming back to Half-Life? Was it a strange feeling revisiting the series after such a long time away from it?

Robin Walker: I don't think it was strange at all. Certainly for myself, and I think Jim probably feels the same way, it was a mix of relief and, err, joy might be a strong [word]... I don't know. I really like working on Half-Life, both [because] of our like of the game and the world and so on, but also as a game developer because of the way we build Half-Life. We build Half-Life in a way that's unlike the other games we work on these days, because it's a very crafted game.

When you build Half-Life, you do it room by room, and we do it with groups of people. Then we go into the next room and say, "All right, so what's novel in this one that we haven't seen in every room prior to this in the game?" and we just keep doing that, and then put them in front of people and playtesters and see what happens and change them and then add more and you just keep doing that until the game's sorta done. In my experience, there's no better way to become a better game developer, because you spend so much of your time just watching people play the game, and that's the secret to how you get better at making video games. So for me, I really enjoyed coming back to it just because of that. I knew I was going to grow more as a developer building another chunk of Half-Life than I was working on another long service game of any kind.

Jim Hughes: This type of product, this story-driven, single-player narrative really plays to the strengths of a lot of people in the company in terms of people iterating on each other's concept and ideas, like basically something that starts out good, you know, somebody suggests something else and it gets increasingly better, and that's happening from week to week. In content like this, a lot of that is basically driven by watching people play, and we've been watching people play this product for, I mean it's a long time now, from when we had five minutes up and running of being able to play in VR, we've been watching people, and then essentially basing our direction next on, well, 'Players seem to like doing this, we need to chase this, we didn't anticipate this.' So in terms of the combination of Half-Life just as a cool space to work in, it's amazing, but at the same time doing it in VR made it extra special.

Watch on YouTube

RPS: During the initial reveal of Half-Life: Alyx, you said that the prospect of making Half-Life 3 was still quite daunting back in 2016, but would you say that making any kind of Half-Life game is still a daunting experience? Or is there something particularly terrifying about putting the number '3' on the end of it?

RW: Ha, we're both laughing!

JH: Terrifying!

RW: I think after the original announcement, I think we all breathed a deep sigh of relief and just got to finishing the game. We used to joke that announcing this game was much scarier than finishing it, which is such a unique experience for us all. But, you know, the reason Half-Life was scary is in some ways really a function of what Half-Life is to us. Like, Half-Life has always been the product where we go in with something we're interested in really tackling, some specific problem we want to solve, or some specific experience or opportunity we can see. If you go all the way back to Half-Life 1, we felt like there was this real opportunity at the time for first-person shooters to be more than corridors and combat. We thought that there was a possibility to do interesting narrative delivery in there, and so Half-Life 1 was very much focused on that opportunity. Not so much on exactly what the story was, but the way in which it could be told in that kind of first-person shooter platform.

Then we go forward to Half-Life 2, which was the place where we thought there were interesting opportunities in how we could do character work, in continuing to try and drive forward how we told stories and narrative in first-person shooters. We also thought there was a real opportunity around physics. At that time, there were a few physics engines just starting to show up and we were very interested in seeing how we could do more than just simulate boxes and barrels, but actually build gameplay out of that.

So I think for a while there after that and after the Episodes, it wasn't so clear what that next thing was going to be, and we weren't really used to trying to build a Half-Life where we weren't quite sure what problem it was solving. Then, I think, VR in this product kinda did that for us. Instead of sitting there at the start of it going, "What is Half-Life 3?" we already had a prototype using Half-Life 2 assets that people could play in VR. It was supposed to be about 15 minutes long, but people would spend about 45 minutes in it, and you'd just watch them and it became really clear what to do. It was like, "I get this." You take all the things that we understand that work in Half-Life 2 and figure out how VR makes them better, and there were some of them that were obvious from day one, and as time went on we found more and more opportunities.

But it meant that instead of sitting there with an existential abstract problem that you don't know how to work on until you've, you know, it's like, "Well, we'll start working on [this] and hopefully in a year or two maybe we'll have got an answer to why people should play this game." With VR, it was like, "All right, I get it, I know what work I can do on this today. I can start writing code, I can start building levels and all this stuff," and so I think that was enough to convince us to, you know, to allow us to dodge that harder existential question and just get to work and do our best there.

RPS: When you eventually decided your next VR game was going to revolve around Half-Life, what came first? Where did you begin?

RW: Technically, the very first rooms were a series of test rooms where there were just Half-Life 2 NPCs. The first thing we did was just to try and see what happens when we try and fight all those existing enemies using our existing weapons in VR. What are the things that are translating well? How does combat change? All that sort of stuff.

But the first thing we actually really built was this thing that we never actually got around to naming, but we just referred to it as The Prototype, which was a short 15 minutes set of six or seven rooms of just zombies and headcrabs. It was largely Half-Life 2 code, it was all Half-Life 2 assets. I think we stole the gloves from Counter-Strike: GO, because they had good glove models, so the only real new code we had was just letting the player move around and have hands to pick stuff up with and drop and throw things. We put a flashlight in at the time and that was enough to start putting people together and let them see. The prototype allowed us to have a shared image in our heads almost immediately, and that was a huge boon, and I don't think we've had that experience so much before.

RPS: With so much distance between now and the end of Half-Life 2, how has your perception of the series changed over time? Both games arguably set the template for every FPS game that came after it - was there a sense that everything that made those games special at the time has now become a bit routine? How do you make Half-Life exciting again?

RW: If anything, certainly after Half-Life 2, I think the industry actually went in a pretty different direction than Half-Life 2. I think more around the time of the Episodes, [we were] entering the era of open world gaming. They're the opposite end of this axis, they're in a completely different space than Half-Life's core DNA of "We're going to craft the heck out of every single room."

So in some ways, for those of us who really like playing Half-Life-style stuff, there haven't been a lot of things like that. Internally at the office, for example, we all love the Metro games because [4A Games] feel like a developer who have taken some of what we were trying to do in Half-Life 2 and have done it better since. So I guess we didn't ever really have that worry. I think if we'd had a religious attitude toward Half-Life and put the games on a pedestal, saying that everything has to be exactly that way, then I do think we'd run the risk of not learning from why the good things worked. We didn't do that. We try to be rational about those products and look at why things were the way they were in them, and use them if they still seemed applicable to this.

RPS: Is there a particular element of Half-Life that you think actually works better in VR?

JH: The exploration was a big deal. When we were watching people play it, they consumed the actual content at a much slower rate, really digging through stuff for items or world fiction. It was really apparent from the beginning that the amount of gameplay mileage we would get out of, if you could arguably say, square-footage of track was going to be way, way higher than what we could have ever anticipated, which was cool. So we tried to densely pack it to match player's expectations. So when we saw players really digging for stuff, it was like, "Well, we've got to give them something to find there," and we ran with that.

RW: I really agree with everything Jim just said there. In general, one way to think about this, is that at the end of the day, the currency we have to spend in the game is the player's attention. Whenever we can align our efforts with the player's desires and experiences, then we're going to spend our development time in the right places.

Half-Life: Alyx playtesters were obsessed with collecting stuff - and we are, too, especially when it's a good-looking bread roll like this one.

Early on we noticed the attention the player paid to the world, and that gave us lots of opportunities to respond to that. I think we had assumed that some other things like combat would be things that wouldn't come along for the ride as well, that we'd lose something there, but we turned out to be wrong on that.

What we found was that combat in VR gave us the best of both worlds. Players actually take longer to do everything, but find the entire experience more involving and more stressful overall, and that gave us all this extra time. So, for example, in the Combine soldier combat, the average lifetime of a Half-Life 2 Combine soldier was probably somewhere between 10-20 seconds. The players really killed them that quickly. But as soon as we started to do playtesting here in VR, we saw enemies lasting for a minute or two of life, and when you're a development team looking at that, that's incredibly exciting because you can do some much more with AI and performance and enemy speech. The number of opportunities you have to create great experiences when you've got an enemy that's going to last a minute or two instead of ten seconds is enormous.

And the best thing is, it wasn't that the players were bored during that extra minute or two. They actually came out of those scenes feeling incredibly stressed, and so some of our pacing had to change. We built, I think, our first three or so Combine soldier fights and fitted them to a pacing we were familiar with from Half-Life 2, and after the first one people were saying they needed a break! So we realised we really needed to rework our pacing here, as this combat is so much more stressful.

JH: But there's a physicality [to it], especially. I mean, it's present throughout the game, but when you have fast-paced combat like when you get to the Combine, you're reloading, you're scrounging for ammo, you're taking physical cover and moving your body, and it's just a different experience. It's very amped up when you're standing behind a pillar and you're physically moving and leaning and you're essentially trying to stay alive. That's a way different experience with the headset on and actually using your body rather than just sitting with WASD controls on your keyboard, so it's definitely toned down. But it felt really cool that everyone came out of [those playtests] with the same reaction of, "Wow, I just felt like I experienced something."

RPS: If VR improves all the things we love about Half-Life, is there now a worry that another traditional mouse and keyboard Half-Life won't live up to the same expectation? In making Half-Life: Alyx, has it actually made Half-Life 3 even more daunting now, or does it seem more manageable?

RW: [laughs] This is the meanest interview ever! We've talked about this and it was just like, "Yeah, why don't we just try and make the best game we can and worry about how much we've screwed ourselves in the future." That seemed like a problem that, if we run into it, we've been successful. Yeah, there's always a point at the end of a product like this where you realise you've been working for, in our case here just about four years on the dot when we release, and that's, you know, four years of making decisions on top of other decisions that the best data we have on whether they were any good has just been from playtesting, and that's one of the reasons why we bring so many people in.

But, you know, it's always a terrifying moment when you hope that somewhere around Year Two you didn't make a bad decision and that pile another enormous amount of decisions on top of that. You just have no real idea in our experience until things are out and "real customers" have it and just play it and see what they think. We're excited because in the next month or so we really get to find out a whole bunch more about all the work and decisions we've made in the last few years, and then hopefully have some idea about what it is we should do next. We try not to... certainly as much as possible, I think, cautiously we try to avoid making decisions that we can put off if we know we'll have more information later. It's common to ask ourselves, "Do we need to make this decision today or can we wait a little bit?" And this is a case where, you know, there's an event horizon coming up right now real soon where beyond it we'll know so much more than we know today, so we're comfortable just waiting for that.

RPS: Half-Life: Alyx supports multiple movement styles. How did that impact the design of the game? How do you account for the time it takes to move and turn in the blink and zoom modes?

RW: Not anywhere near as much as we thought. We thought when we started the project that Blink and Shift, which from a gameplay perspective are pretty much identical, and Continuous were going to be really different things, that we'd have to do a lot of custom [stuff] that would have an impact on the game design outside just the code you need for the two versions. But it just didn't end up that way.

One of the things we noticed around teleport that was really exciting to us was that it tends to be jarring when you watch someone else do it, but when you're in it yourself, you don't even really notice it. In fact, there's more nuance to it than that. We found very clearly that people only noticed it when they were in worlds that didn't hold their attention. So when we had our early maps where there would be grey boxes and we were just testing and there was no real detail to the world, there was a period of time when we had a second map through the Quarantine door [in Chapter 2] had a bunch of art in it. The map prior, the entrance map, was still grey-boxed. We'd bring playtesters in and they would talk about teleport and what they thought about it in that first map, and then the moment they hit the next map where all the art showed up, it was like they'd hit a wall. They would just stop talking entirely about teleport and just focus. It was such a strong reaction and so consistent that it caused us to try and spend some time trying to figure out what the change was there. And we came up with a series of answers about density and so on, that we should avoid areas where you ever really need to teleport more than a couple of times in a row, and we should reduce the amount that you need to teleport.

Things like the gravity gloves let you walk into a room and pick up three things from a distance – you don't need to go to each of them. But over time, the heart of what we came to realise was that once players were engrossed, players just kind of edited that movement out. That was a combination of both the density but also the experience they had with the actual teleport themselves. Whenever you acquire a new skill, there's a lot of focus about the execution of that skill. If you think about reloading, you're still paying attention to your hands and looking at the gun, because it's still a novel enough thing.

Once you get used to it, the execution of it falls away and you become more goal-oriented. You just think, "I need to reload" and you do the reload and you'll still be able to keep an eye on what enemies are doing and make other choices. And we think that movement and teleport became the same way. When you're first start with it, you're learning to orient and position yourself in the way you'd like, you'd pay attention to it, but as you get more experienced with it, you start to abstract that away and become more goal-oriented. So I think, in our experience, we felt quite comfortable that players, whichever movement choice they make, they all reached the same point after a couple of hours into the game where they've stopped realising what they're doing movement wise and they're just playing the game.

RPS: At what point did the gravity gloves enter the picture? Were they a real early thing you settled on or did they come a bit later?

JH: They were pretty early on. I can remember some of the early things that weren't planned, like being able to gravity-pull ammo off a zombie that was coming at you – in other words, in order to reload your gun, you had to pull it off the guy coming at you. I can remember that, like, actually earlier on, but they were in there more as a way to easily facilitate the easy picking up of objects without having to physically bend over or do something like that. It just became a natural extension of how you interact with the world. They got really fine-tuned in how they choose to select objects – there's a lot of smart stuff going on under the hood in terms of how they operate – but early on, from the beginning, they were in pretty early on from what I can remember.

RPS: Gordon Freeman has obviously always been a silent protagonist, but here Alyx is constantly chatting with Russell over her headset. How did having a speaking protagonist come about? Was it simply because Alyx has always been a speaking character and having her suddenly go silent would be weird?

RW: All of the above, I guess! This is perhaps a good example of what I mentioned earlier around how we tried to treat the old games as documents containing a bunch of good game design decisions for those products that we should think about and ask ourselves about their applicability to this product, but not just adopt them blindly or wholesale or anything. We spent a bunch of time early on messing about with Alyx, trying to figure out what kind of language players responded well to, where it was appropriate, etc. There are moments early on when she did more emoting around things that were more aligned with the player's experience, and there were pros and cons to all of that. We did a lot of experimentation early on, but over time there were certain things that became sort of obvious that we needed to do.

A couple of points, for example, when we'd ask playtesters questions around what they were doing, the state of the world, their goal etc, it was interesting how, when Alyx was silent, they just assumed they were playing as Gordon Freeman. Another part was that we were never religious about, even in Half-Life 1 and 2, that the idea of a silent protagonist was the only way a game should be made. It made sense for Half-Life 1 and 2 and what we were trying to do with narrative in those games, but in this one it seemed more rational and more logical as to why Alyx would talk than not talk. In this case, you're playing a character who does exist. When you started Half-Life 1, you didn't know who Gordon was or a sense of his personality or anything, so it was easier to align him with the player as much as possible. In this game, you know who Alyx is, she has a personality and so on, so it makes sense for her to talk.

And then there were later points in the game when we thought it was important to be able to have conversations between Alyx and someone else because of other problems we were running into. We were very cognisant about trying to make a game that we wanted as many people to get through as possible, while still at the same time remaining true to Half-Life, which meant that we had to have some horror in it. We have a bunch of people on the team that don't do well with horror games, myself included, and we know that VR will multiply all that tenfold. So we were very cognisant of, how do we help players get through this, and Alyx being able to talk with Russell was an important tool for the pacing around that. Sometimes we need to be able to give you an emotional breather, but without having to get you up to open daylight somewhere or something, so one of the tools was being able to have Alyx and Russell have a conversation about what just happened and so on. Just to keep that emotional pacing we needed.

RPS: On that note, you've now got headcrabs literally jumping at your face. Was there a discussion about whether they should jump directly at you?

RW: [laughs] There are two schools of thought in the company. One is that we should never make a Half-Life game in VR because of headcrabs jumping at your face. And the other camp is, "We should definitely make a Half-Life game in VR because of headcrabs jumping at your face." We do do some stuff – there's actually a percentage chance they have to deliberately miss, so they'll leap to the left or right of you instead of at you.

That wasn't so much so we could avoid them jumping at you, but it actually turned out just to be more fun. Players tended to be really good, just in that first moment of the leap, seeing the direction the headcrab was going and suddenly moving to the side and feeling like they'd dodged the headcrab. That's the sort of thing that, at the end of the day, we felt like we couldn’t cut – it was too Half-Life and we had to instead consider that a constraint and figure out how to do that. We've attempted to be as fair as we possibly could in the early hours of the game. There's a lot of careful design in the enemies in terms of their ability to look like they're threatening you, but not really being able to threaten you and give you time to deal with things and move forward at the speed that you need to be able to handle what we're throwing at you. So we're very careful with that stuff. Chapter 2 [where you first start encountering enemies] is the most playtested area of the whole product by far.

RPS: You talked earlier about just how much time playtesters were spending looking around these environments and that you played to that strength. Would the game's upgrade system not exist if players hadn't spent so long looking for things?

JH: One of the funny things about the upgrade system was in the beginning we hadn't done the work for the actual upgrades, but we'd put in the collection resin, and people didn't know what it was, and they were still collecting it obsessively, saying, "This is something important, I want this!" So that was successful. We joked around saying, "Let's just make them collect it and then they never get to do anything with it!" But basically, people were obsessed with collecting stuff in the world before they even knew what to do with it. But the upgrade system was always the plan.

RW: I remember there was a playtester whose feedback afterwards was, "I really liked collecting that stuff. I don't know what it's for, but I really liked collecting it!" But yeah, to some extent, I think Half-Life 2 informed that. We went in with that idea already because there were periods of Half-Life 2 where we really struggled to create optional stuff, like there were often points in a Half-Life level where you'd want to build some side areas, and all we could ever give you in Half-Life 2 was ammo. So if you were a good player, you were never starved for ammo. We would try and put some novel experience around those side areas, but at the same time, players often leave those areas wondering, "Did I come out of that with more stuff than I went in with?" And because the early prototype had shown us that people wanted to explore the world, as Jim says, we were already thinking about how we were going to reward them for that. We needed something you're always going to want, no matter how well you've done so far, and that's where all that started from.

RPS: Is it harder to tell a story in VR? You can't direct a player's gaze as easily as you can in a traditional first-person game and you can't lean on in-game cutscenes. Did this present new challenges for the team?

RW: If anything, it's no harder than Half-Life 2. In Half-Life 2 you'd watch a scene playing out, and every tenth playtester in Kleiner's lab would be jumping on his head and climbing up into the rafters. If anything, it was actually in some ways easier in VR because players do look around a lot more. One of the problems you always run into in an FPS is that players are really good at looking horizontally with their mouse, but they don't look up and down nearly as much. You've really got to work to make a player look up in FPS games.

But in VR, we found that players looked up and down a lot more than in any of our previous games, and so there's a whole grab bag of tricks to get you to look at something, and a lot of those were the same things we used in the past, it was just that this time around they were even more effective. There's a lot more narrative in this game than any of our previous Half-Life games. Some of it is due to the density thing Jim talked about earlier. There are things to find and look at with a lot of story on them. Because we know you can actually look at them closely – like with a newspaper, we don't just blur out the words, it's got everything written on it – and you can talk to Russell about them and there are threads of conversation you have throughout the game that rely on you as a player having found things as you go through, and stuff like that. All those things were opportunities we had because players paid so much attention.

RPS: Was there anything you learned or took from the past few years of first-person VR games made by other companies? Are there VR FPS games that you think do a great job? Are you defining the standards for yourself or have other games already done it?

RW: There's been a huge amount of awesome VR stuff built over the last few years, and we looked at a whole ton of that. I mean, we cheated right? We have a much larger set of resources than a lot of the other folks get to have. In a lot of ways, really, we thought that that was our role, in terms of helping to push VR forward. There were a lot of people who, in that first year or so – we started this shortly after that first year of VR – there was an enormous amount of super creative work done to explore what kind of mechanics work in VR and what mechanics don't, and in a lot of cases, that work was done by very small dev teams who could explore a mechanic heavily, but weren't necessarily able to invest heavily in building a large-scale gaming experience around that mechanic. So we felt that was our job, since we could do that, was to really build that large-scale experience around the mechanics.

RPS: Half-Life Alyx is playable on practically every headset under the sun, but it's been made with the Index in mind. Why did you feel the need to build your own headset?

RW: To some extent, this goes back to why we got involved in VR in the first place. Back before the original Vive, when we were first looking at VR and starting to think about it, there were a bunch of people interested in VR, or a bunch of the people who I think looked like they were going to push heavily on hardware had ideas for what VR should be that worried us, I guess. We felt that they were going to really limit the opportunities for it as a medium, and their focus was much more around looking at it as a viewing medium. Their model was like VR was having a screen really close to your eyes and that was the difference. And we were really worried that it would be a thing that we would lose an opportunity for building some really interesting experiences.

So we felt like game developers had to get involved to make sure that, as a medium, it actually fulfilled its promise in terms of what it could be. In particular, at that time, we thought very heavily that input mattered a lot. We didn't want a screen this close to your head and then have a controller that looked like a remote control, or anything that only allowed us to control media on it, or something like that. As game designers, we felt that track control was having two hands, and that was much more empowering game design change than just having an HMD on. So we got involved because we thought we had to. We worried that if we didn't, no one else might.

And then, Index comes along later because we continued down that path. There are still a set of things we thought were really important. In the early days, we started testing this product on older headsets and then moved to the Index just a year in or so, and we really got to see what that transition gave us. There were a bunch of things I remember having to explain to playtesters, like how to pick something up because it wasn't necessarily obvious to people who hadn't played a lot of games before, but the day we started playtesting with Vives, we stopped having to tell people that because it was just so natural. As we worked on it, we were able to build a platform now that is approachable to a bunch of people who in the past wouldn't play games. My mother could play Alyx. She would never be able to play Half-Life 2. At the same time, though, we didn't make any concessions in that way. An experienced, hardcore player has far more power to control their world in Alyx than they did with keyboard and mouse before, so it's the best of both worlds.

Other things like making [hardware] so you could play a three or four hour session in VR without being physically fatigued was a big part of Index as well. The fact you can rest your hands without dropping your controllers was just enormous. Our playtesters went from being able to do an average of 45 minutes, which was sort of the rule of thumb based on all the data we had when we started this project, to doing 3-4 hour sessions, breaking for lunch and then coming back and doing around 3-4 hours because they could. And that was down to things like the controllers, and the headphones off your ears. All these things came together to mean you could really spend a lot of time in there.

RPS: So now that you've made 'the big VR game', what's your take on VR as a whole? Where does it go from here? Are there still hurdles that need to be overcome?

RW: I think there were always a lot of venture capitalists who really wanted VR to explode in some way that we'd never seen hardware do before, but VR adoption has gone pretty much the way we thought it was going to. We've always said we should be realistic about how this is going to go. I think that we leave this now convinced that there's a really interesting medium that can build experiences that we haven't been able to build on our previous platforms. We would never have believed that VR was going to replace anything – that's all silly talk.

The only real thing that was interesting to us was, "Is it a novel enough medium? Does it have its own strengths that we can build something in it that is unique to the medium that is really interesting to a whole bunch of people, and that they leave saying, 'Yes, that was cool, I want more of it.'" They can keep playing their PC games as well and that's all great as well. We don't want to tell them they should stop. But, you know, for us going into this, that was essentially the question – how unique and novel a thing can we build here, and at the end of it is it a thing that enough people are going to want? I think, at the end of Alyx now, we think we know the answer to that. In a week, we'll find out if we're right, so I think we just hope that a lot of people get the opportunity now to play it in VR and see why we chose to do this. I think that once they've done that, we hope and believe that they will understand after a couple of hours like, "Oh I get why they did this," and that would be cool!

Read this next