At the annual Game Developers Conference this afternoon, Google finally unveiled Stadia, its video-game program. It is both technically impressive and a little depressing. Stadia is a hardware-agnostic game-streaming platform; the software runs on a remote server and sends the video feed to the end user, which means that the only thing that user really needs is a display and a controller. Late last year, Google ran a beta program called Project Stream, which allowed players to test out a copy of Assassin’s Creed Odyssey running in the Chrome browser. It worked better than you might expect an interactive stream to — high-fidelity visuals, low-latency inputs — but then again, Google has more muscle than most of its competitors in pretty much every tech area.
Much of Google’s presentation was showing off that muscle. Stadia is running on a powerful hardware stack — 16 gigabytes of RAM, a custom GPU capable of 10.7 teraflops, a “2.7 GHz hyperthreaded x86 CPU with AVX2 SIMD and 9.5 MB L2+L3 cache,” if any of that matters to you (dork) — and it is being built with support for popular game-developer tools like Unreal Engine and Unity, if that matters to you (nerd). The next Doom game, in addition to launching on traditional home consoles like PlayStations and Xboxes, will also be available through Stadia. The point of much of the presentation was that developers could put their games on Stadia with minimal effort, immediately adding a third competitor to a game-streaming race already occupied by Sony’s PlayStation Now service and Microsoft’s nascent Project xCloud.
The lone piece of consumer hardware that Google unveiled was that Stadia controller, which looks like your standard controller (instead of connecting to a console, it connects directly to Wi-Fi). Like the controllers for the PlayStation 4 and Nintendo Switch, it has a “capture” button that can record gameplay and, in this case, broadcast it to YouTube. The lone differentiating function is that it has a microphone onboard and is integrated with Google’s Assistant for various unspecified, developer-integrated features.
The more intriguing aspect of the presentation, however, was Google’s heavy emphasis on Stadia integrating with online social spaces. The goal is to “combine the world of people who play games and people who watch games into one global community,” longtime game-industry exec and Stadia head Phil Harrison said at the end of the keynote. On the simplest level, implementing this goal means that the video stream Google’s system sends to a player can be mirrored on their YouTube channel instantly.
But there are other features that go beyond simple streaming efforts. Google showed off Stream Connect, a feature that merged different Stadia feeds into a single view, meant to emulate the days where people sat on the same couch looking at the same TV when playing split-screen multiplayer games. There was also a feature called Crowd Play, which let stream viewers enter a queue so that they could instantly play against a streamer they’re watching. The Google Assistant button on the controller could maybe be used to call up a video tutorial for how to get through a level or solve a puzzle.
Perhaps the most intriguing, forward-thinking feature is one called “State Share.” The idea behind it is that Stadia would allow players to share their game saves with others — aspects like the state of the in-game world, the player’s in-game location, or even the items in their inventory. Those values could then be hashed into a URL and shared with friends or widely, across any social-media platform that allows for hyperlinks. A hypothetical example: If a streamer, low on ammo, managed to pull off a near-impossible victory, they could then share a link which would allow viewers to attempt the same challenge instantly.
As with all developer conference presentations, you should not expect Stadia to explode out of the gate. It’s supposed to launch this year, but whether or not developers will make their games available on Google’s system has yet to be seen, and whether they will adopt Stadia-specific techniques is even more up in the air (Google does have an in-house studio working on games as well, however).
Not that I want to cast doubts on Google’s commitment to gamers, but it’s difficult to not see the Stadia push as an enormous subsidiary of and supplement to YouTube, the perennial runner-up to the Amazon-owned Twitch. Some anecdotal stats I recorded this afternoon: Fortnite had 159,000 viewers watching live on Twitch and 46,000 on YouTube; League of Legends had 143,000 on Twitch and 12,000 on YouTube; Apex Legends had 272,000 viewers on Twitch and 52,000 on YouTube.
Baking a gaming service directly into YouTube has a pretty good chance of increasing the engagement metrics that power the platform’s ad service, and creates a line for direct purchases. It’s gaming with almost no friction: no waiting for games to download, no waiting for friends to log on, no calibrating one’s streaming setup, no leaving the app to find a video tutorial. At one point in the keynote, YouTuber MatPat explained to the developers in attendance that “creating challenges through State Share allows us to continue creating content for all your favorite games, long after the game itself has been completed.” How … inspiring?
The looming issue of linking Stadia so closely with YouTube is that it creates a feedback loop of engagement metrics that ends up predetermining user behavior, and maybe that user behavior ends up influencing game-design choices. YouTube’s recommendation algorithm is, by most accounts, the worst feedback loop imaginable. It’s not hard to imagine a world in which the Stadia platform ends up wielding major, indirect influence over what types of things game developers create and what YouTubers decide to stream.
Earlier this month, Kerry Davies of the analytics firm Fancensus revealed a near-complete lack of variety in which games that streaming influencers covered. In 2016, the top ten influencers covered 383 different games on their streams. In 2018, the top ten influencers covered only 28 games — 28! “This is a clear indication that influencers are focusing on what drives the most views and/or the relationships they are forging with major publishers, with half of those 28 selected games being AAA console titles,” Davies wrote.
While Stadia might, on a technical level, allow more game developers to make their games accessible to a wider audience, the corporate strategies of social tech platforms like YouTube and Twitch frequently encourage and reward homogeny. In other words, everyone’s playing the most popular thing and chasing what’s trending — and that philosophy could easily seep into the heart of Stadia, if it hasn’t already. Should we really be optimistic about a platform so closely linked to the core mechanisms of YouTube?
It is difficult to look at Stadia’s proposition without thinking about what YouTube and other content sites like Netflix have become. YouTube often rewards imitation and videos about trending topics (a consistent headache for the site) and Netflix’s original programming is, by and large, mediocre comfort food — shows that feel like they were generated by a matching algorithm instead of creative humans. The human choices are driven by algorithmic results. It was all I could think about when Google’s Erin Hoffman-John showed off a game-development tool called “Style Transfer Machine Learning,” which allows developers to use still images to immediately create a visual palette for a video game, making prototyping and visualization faster and easier. “What excites us and some of our partners so far is how many playable art styles we can create just by feeding it more and more images,” she explained. Quite literally, it was an algorithmically generated video game, created by shoveling more and more data into the Google beast’s maw.