Valve on Episode One

Whether by design or simply by default, Valve is leading the charge on episodic gaming at the moment. For all the talk that's been generated about this kind of segmented gaming, it is likely that most developers and publishers at this point are keeping their eyes on Valve and the few other studios trying out the episodic model before they themselves jump in. Recently, I visited the Valve offices to play through the company's latest release, Half-Life 2: Episode One, and I was asked what other developers have been impressing in the episodic arena. It was a rather difficult question to answer. There's Ritual with SiN Episodes of course, using Valve's own Steam service for distribution, and there's Telltale with its Bone and upcoming Sam & Max adventure game series. Out of the very few companies taking the plunge in the early stages of the format, Valve is the one which has most frequently spoke out in favor and defense of episodic gaming--or, at least, the one which has most frequently been asked about it. So I asked some guys from Valve as well, and they answered. Why episodes? Why not keep going with the proven method? Both of Valve's two main Half-Life titles have been praised as some of the most immersive and revolutionary shooters ever made, and each game is a rather lengthy epic. There are actually quite a few reasons for the change. Having only just released its first piece of episodic content, this type of game creation is still new to Valve. However, the company is fully committed to the method. "We are definitely doing this instead of Half-Life 3 right now," said Valve's Robin Walker. As it turns out, episodic delivery is actually incredibly well suited to the company. Some of the reasons are quite straightforward. "There's a lot of depressing evidence out there indicating that not very many players are finishing out games," admitted Walker. "As a creator, you want people to see all the cool stuff you've made." This is something that plagues video games as an entertainment form perhaps more than either the players or the developers would usually like to admit. It may not be as much of a problem among core gamers, and of course those who are fans of a particular game or franchise are much more likely to play to the end. Unfortunately, that's simply not true for most gamers. As much as people cry out for lengthy game experiences, most of the time that length is not being taken advantage of when we get it. For the record, when somebody at Valve talks about "evidence" relating to player experiences, it's not referring to extrapolation from message boards, or volunteer surveys. "Steam allows us to see where people gave up," Walker explained. Steam Steam, while primarily acting as a content delivery system serving Valve as well as a growing number of other independent developers, is also a tool Valve uses to guage gamers' pace, gameplay tendencies, and success rates long after the company's already extensive internal playtesting has finished. The episodic format is what allows the company to actually take advantage of that data in a timely fashion. Walker compared Valve's interpretations of Steam data to the process taken to create and polish mods such as Counter-Strike or Garry's Mod: "Episodic helps gameplay iterations. We want to build a way where we can make a single-player game with those advantages." For the record, Valve co-founder Gabe Newell emphasized that Valve hopes players know what data is being collected and why. "We're paranoid internet users too," he said. "The hardware data we collect is published online." That data is used to give Valve, and other publishers for that matter, a good idea of what kind of PC hardware actual gamers are using, which helps developers know what to shoot for in their games. Gameplay data is used internally to refine gameplay in future releases. Newell hopes that such refinement can be taken to even greater levels in the future. He proposed the idea of a multiplayer game that could, for example, balance weapon prices in real time based on the in-game equipment purchasing and usage habits of players. How Valve makes games As you probably know if you've listened to Episode One's commentary tracks, Valve has a very iterative development process. Rather than designing a large level and then populating it with enemies and objects and perhaps some interesting scripted events, Valve focuses on specific encounters. Designers decide what needs to be conveyed in a particular encounter, discuss how to convey it, and test the implementation again and again and again until it works. Because Valve's storytelling have some very specific constraints--the first person perspective is never broken, the player never loses control without good reason, the protagonist never speaks, and so on--this development method is almost necessary to ensure that the intended story and emotions are reaching the player intact. "Every line of dialogue should reveal something about the world," Walker stated. "We want new questions raised in each episode, with enough for people to generate discussion." Newell chimed in, "We're all big J. J. Abrams fans here," referring to the creator of television programs such as Alias and Lost. This technique benefits from episodic production in two ways. First, when Valve gets player feedback on what players think what works and what doesn't work, the company has the opportunity to iterate on gameplay using that information in a matter of months rather than a matter of years. For example, Valve got excellent feedback on the commentary feature in Half-Life 2: The Lost Coast, so it was carried over and expanded for Episode One. "We had fourteen pieces in Lost Coast; in Episode One we have about 115," said Walker. Continue on to page 2 for more on Valve's development process. _PAGE_BREAK_ When it comes to pacing, there's also a less obvious benefit simply in the shorter length of an episode. As great as Half-Life 2 is, Valve's design process sometimes shone through a bit too much. Though the game is not divided into discrete levels as most first person shooters are, the game sometimes felt segmented regardless. There were many different types of gameplay, but those segments were perhaps more clearly defined than would have been ideal. When the game is only several hours long but still has a good variety of gameplay situations and attention to detail, as any Valve release does, everything becomes denser and more tightly constructed. Plus, any situations that still don't work are more likely to be corrected without having to wait half a decade. Walker gave one example of how this impacted Episode One. "People really enjoyed the super gravity gun in Half-Life 2," he said, "so we gave it to them almost at the start in Episode One." To go along with that, Valve increased the overall scale and depth of interaction with the game world. Speaking of Valve's development process, I was curious as to who actually does what on these games. It turns out nobody was really able to answer my question, or at least nobody wanted to. "Our business cards say all kinds of things," laughed Newell, who himself wears many hats at the company. Obviously, certain people do certain things--there are animators and writers and artists and so on--but everybody working on a given project has a high level of input into the decisions made on that project. This is evidenced by the way Valve does credit sequences: all of the team members are listed alphabetically without any titles. Initially, I tried to find out who acted as "lead designer" on Episode One. When pressed, a reluctant answer was given that Robin Walker is the "project lead," but he was quick to insist that the informal title relates more to general team management than creative control. This too drew a few knowing laughs. "If anyone tried to say, 'I have a vision,' they'd get mocked out of a meeting," explained Newell. "You bring in one wrong person to a team, and it would go off-kilter real fast. Our ability to do what we do is dependent on that dynamic." Another benefit of working with that method, Walker explained, is that since any proposed additions are considered and discussed as a team, it can prevent feature bloat. "It's really easy to add features to your game that actually hurt the core experience," he said. As an example, he described a minor "feature" from the development of Half-Life 2. At one point, players could do a small amount of damage to zombies using just the gravity gun, causing a visible physical reaction from the zombie. When players discovered this ability, developers noticed that they frequently ignored other, better, methods to take out the zombies, such as the ability to use the gravity gun to launch a circular sawblade at the enemies. The damage and visual feedback was nice touch, but it actually detracted from the game at the time by slowing down players and weakening the pacing with an inefficient combat method. Tech benefits Valve doesn't just make games, the company also maintains and licenses its Source engine, used for all internal projects as well as numerous games from other developers. Working episodically, Valve can not only update the engine on a frequent basis, but can also ensure that all of the significant improvements have actually been demonstrated in a shipped product. This is helpful to licensees because it provides an immediate showcase for new features. For Episode One, the second major version of Valve's facial animation system was released. High dynamic range rendering (HDR), first showcased in Half-Life 2: The Lost Coast and Day of Defeat: Source, was improved and put to great effect in Episode One. The engine has also been optimized a great deal. Walker said that Valve learned some optimization tricks through the process of porting Half-Life 2 to Xbox, and many of those techniques have been applied to the Source engine on PC. "We pushed forward with graphics," he explained, "but the low end requirements got even lower due to optimizations." One interesting example is that Valve noticed, quite frequently, stability problems were greatly reduced when users defragmented their hard drives. Taking that into account, Steam now automatically defragments the game cache, which has the same effect. Risk management One important benefit of episodic development Gabe Newell mentioned is risk management. As development costs, both on the PC and on consoles, continue to increase, simply developing a video game at all is a substantial risk, so most publishers try to find external factors such as licenses to minimize that risk. "When making a PlayStation 3 game, you've already expended all your risk," said Newell. With smaller scale episodic game, developers are putting less on the line with each release, so there is more room to try new concepts and approaches. In regards to consoles, Valve has announced it will be supporting Xbox 360 in some form, but Newell had some positive remarks about Nintendo's approach as well. "I understand why [Nintendo is] doing these things," he said. "I think they'll see more success than just making PS2 games with four times the budget and a movie license and a big day to date tie-in release." To wrap things up So then, it's probably just as well that everybody is looking to Valve to gauge the success of episodic gaming. After all, Valve seems to have the most answers when it comes to the subject. Early reports indicate that Episode One is doing nicely in retail, not just online; Valve announced today that the retail version of Half-Life 2: Episode One debuted at #1 on the European retail PC sales charts upon its release. While there are still only very few developers working on these kinds of projects, momentum is undoubtedly growing. Both Ritual and Telltale Games have expressed interest in bringing their episodic projects to consoles, which are starting to see digital distribution for the first time. Personally, I'm looking forward to actually finishing more games.