=========================================================== By Ben Lang Published April 14, 2020 https://www.roadtovr.com/valve-half-life-alyx-game-design-interview-robin-walker/ =========================================================== The secret to Half-Life: Alyx’s near-universal acclaim is… well, not as secret as you might think. Yes, the game surely had a budget much larger than any other VR game to date, but it’s far more than money that makes a great game. In an interview with Half-Life: Alyx game designer Robin Walker, I got a glimpse of Valve’s game design process which felt intriguingly different from much of what I’ve seen elsewhere in VR game development. As Walker put it, creating Half-Life: Alyx was truly approached ‘one room at a time’. --- The way we build Half-Life is unlike how we build our other games. The other games are sort of a service-multiplayer games […]. When we build Half-Life, we just build it one piece at a time… conceptually one ‘room’ at a time. And for each room a group of people sit there and they think: ‘what happens in this room that hasn’t happened in any of the previous rooms and fits into where we’re going with the next room?’ And then once you’re finished that process and you’re happy with it, you put it in front of some playtesters and see what happens, and you iterate on it, and then you go onto the next room. And you just do that until you’ve built the whole game. It’s almost an ‘easy’ process… it’s a very fun process, because it involves a huge amount of just watching people repeatedly play the stuff you made—which was way more fun when working in VR than anything in previous games. --- So rather than approaching the game’s structure from the top-down, starting with high-level gameplay concepts, as many seem to do, Valve really honed in on the gameplay of each room in the game, almost like creating a string of mini-games that were each designed to be interesting and fun. Crucially, this allowed the studio to regularly playtest discrete sections of the game—and then remix, rearrange, or even cut them as needed. Putting playtesters through each room regularly also allowed the developers to anticipate and react to what future players would do in instances where they found that many playtesters behaved the same; walker explains how watching playtesters helped Valve decide where to focus its development efforts in each room. --- The thing that [this process of building rooms and playtesting] really allows you is to respond to what you’re seeing players do. I like to think of it like: you’ve got a [conceptual] ‘budget’ as a developer and the real choice is where you spend your ‘budget’, and your goal is to spend it in the place that generates the most value for the most players. And the great thing about VR is that everyone seemed to react so similarly [in playtests] […]. Where in Half-Life 2 we might build a room—and there might be a vent on one side and some stuff on a bench or something—maybe 75% or 80% of players run through the room [without looking at anything] and 5 to 10% of the players explore one thing and another 5% of players explore another thing. [When only 5% of players are looking at those things] it’s hard to justify spending a little bit of time going ‘let’s give them a unique little piece of content or experience if they explore that thing’. But when we watched [players in] VR early on it was like ‘oh my god, everyone looks at the vent, everyone pokes their head under the desk’—then it became ‘oh great, now we can afford to put all that budget there’. --- Among other things, many of the game’s contextual voice-lines were added only after seeing that many players would do (or even say) similar things as they played. Seeing that common behavior helped the developers decide where to spend their finite resources. Walker said that this process—of testing, iterating, and testing again—created a feedback loop which explains why Valve decided to focus so much on extremely detailed environmental art. --- There were areas of the game that were essentially art-passed fully multiple times. The section right after you go into the quarantine zone was the first area of the game we really built because it was [essentially the game’s tutorial]. We built it first and we tested it before it had any art, and then we put art in it, and then players explored more [because of the extra art], so we put more art in it, and then they explored more, so we put more art in it—and it was just sort of comical how much… you could do that forever it seems […] --- Had Valve not observed that players responded positively as they enhanced the artwork, the studio would have spent far less time on that aspect. For Valve, constantly running playtesters through the game was so critical to the game’s development process that the developers built tools to aggregate player data so that it could be reviewed to answer key questions about the game, like how much of the game’s useful objects (ammo, resin, health, etc) were players actually finding as they progressed. --- Everything you find [in the game] has been hand-placed and chosen as to what it is. We’ve done multiple complete passes over the game about exactly what you find and where you find it. We stat-gather the heck out of it—we have a whole system for seeing… [for instance we would] gather the data from 10 playtesters and walk through the game linearly and see ‘alright, 10% of them picked up that one, and no one found that one, and these three people found that one…’ and be able to really track it. So we did a lot more work on that than we’ve ever done before on any of our games to try to make that payoff; we have a specific experience we want to deliver in that—we want you to always feel like you have ammo, but never as much as you’d like. And that’s true of all the other things as well whether it’s health or resin, etc. --- The deceptively simple part about building a game one room at a time is the seeming inelegance and constant playtesting throughout the design process. It’s the opposite of building high-level systems—like procedural planets, weapon archetypes, or dynamic daily objectives—and hoping for interesting gameplay to emerge. It’s the brute force approach to creating fun. Build a room. Test it. Is it fun? No? Fix it. Is it fun yet? No? Keeping testing and keep working on it until it is. And if it isn’t? Throw it out. Finally made a fun room? Time to make the next one.