We're Not Ready for Streaming Games

Recent events have drawn a great deal of attention to the fight between GameStop and OnLive. Both companies are clearly willing to take extreme measures to dominate the emerging field of game streaming, and all of us on the sidelines are forced to ask: Who will emerge as the top provider of this new medium? Speaking only for myself though, I would suggest that it’s a moot question; GameStop and OnLive are engaged in a race to see who can jump the gun first. Game streaming, in its current form and in this current environment, is not feasible for the mass market.

This is not to discount the efforts made by companies such as these two and others; I expect that we are going to learn a great deal from these early experiments. Early adopters will help to point out the challenges that game streaming must face as it moves along the road to becoming a reasonable option for the average gamer. Before it can get there, however, it’s going to have to make a number of stops along the way, and learn to overcome hurdles related to technology, business models, and the games themselves.

It’s easy to forget that streaming media is not a one-size-fits-all delivery mechanism. At a time when services like Netflix and Hulu are enjoying tremendous popularity with their streaming options for movies and television, it’s natural to follow a train of thought that envisions an equally successful service for video games. Bear in mind, however, that these services are providing you with a static, fixed product, allowing you to view a video that has already been recorded, converted into a compressed digital format, and is linear by nature. The stream works because, when all is said and done, you’re simply downloading a video file – you just get to start watching it before the download has finished.

An online video game does not have the same features as a pre-recorded movie or television show. The picture and sound are, by definition, interactive, and will need to change in real-time based on a player’s input. The closest analog to this that we’ve seen is IPTV, or live television via internet streaming, and even that is a process that has a long way to go before it can match the quality and consistency of a dedicated broadcast. It doesn’t take more than a small burst of latency or a brief connection hiccup to remind you that you’re watching a program on the Internet, and while this might manifest as a mere annoyance for a television program, for a fast-paced, reflex-based video game, it may spell the difference between life and death.

For that matter, streaming doesn’t just need to master the difference between fixed video and live video games: the technology must be equipped to handle a wide variety of game styles and genres, each with their own demands. Video streams maintain their quality by building up a buffer of data ahead of what’s currently on display, allowing the feed to continue during moments of latency. Some games might allow this, depending on how linear they are. For example, if a game knows that the player will assuredly be playing levels in a fixed order, then Level 2 can start downloading in the background while the player is still progressing through Level 1. What happens, though, when there is an open-world design? In a non-linear game, the player may opt to explore any portion of the game world at any time, preventing the system from predicting which portion of the game to pre-load.

Thus, with the data necessarily needing to be transmitted to the player in an ongoing real-time fashion, the problems of latency become inevitable. How can a streaming medium keep up with the requirements of today’s high-end games? A slower, more ponderous game might be feasible, such as a turn-based strategy or card game. Anything demanding precision, reflexes, or rhythm, however, would be extremely frustrating to play. The system can’t rely on Hulu-style “buffering” pauses during a deathmatch; the first time a player is killed while his game catches up to real-time is likely to be his last time playing.

Another popular method that video streams use to keep up with a fluctuating connection is “scaling,” or “adaptive streaming.” If the stream can’t maintain enough bandwidth to keep the show going in full HD display, the bitrate is dropped to a lower quality for the sake of a smoother, uninterrupted presentation. Decreasing image and sound quality would seem to fly in the face of modern game development, with studios continuing to push the envelope to deliver better graphics, more fluid animations, and rich, cinematic experiences in their triple-A titles. A visual benchmark like Uncharted 3 or Battlefield 3 is the product of years of hard work by incredible artists and incredible programmers. Why should a studio spend months of hard work and millions of dollars to create a pixel-perfect, life-like game world, if the end user is only going to see it as a blurry, stuttering mess? For developers to continue exploring what is possible in games, they need to be assured that they can present the optimal experience to players, and streaming games can’t promise that yet.

As I’ve cited in my previous blog posts, the average speed of broadband internet fluctuates across different regions, even within the United States. Our recent study found that the US average was 5.5 Mbps, but it is crucial to remember that that figure is the average. Much of the country has to deal with speeds less than that, and as we’ve just established, there is not a single “ideal” data rate that will be adequate for games of every type. If a game happens to need an above-average speed, a majority of gamers won’t be able to play at a reasonable quality.

Lastly, a discussion of the streaming business model is in order. With the cloud unready to accommodate the average gamer, publishers may realize that streaming is still a ways off from being profitable. For starters, the entire concept is still in its infancy, so there aren’t any established business models or case studies to demonstrate the expected ROI. While a streaming game service should generate increased game play and subsequently more revenue, what are the added costs to publishers? One can assume that, in order to pay for hosting, infrastructure, etc., a streaming provider would need to charge publishers based on some measure of usage, i.e., a company would be charged more to stream games that are played for longer periods of time.

This throws into question the pricing model for games distributed online. Titles are often lauded for the tremendous amount of entertainment they can provide, where an epic-length RPG or a highly replayable shooter can last a gamer for dozens, even hundreds, of hours. When delivered via a streaming service, however, suddenly that advantage comes at a much heavier cost. Publishers of traditional games may have to attempt to sell in-game goods or charge for previously free content in order to drive up RPU and offset this new expense. While this may sound similar to the business model already employed by other games (such as free-to-play titles), those will also face increased revenue demands as publishers will be paying not only for the infrastructure to connect players to the server, but also the computing power to operate the game clients on their own hardware, rather than the users’.

There’s no doubt in my mind that, as technology and infrastructure continue to improve, we will one day see streaming games as not just a reality, but as one of the default ways we receive our games. Just as we’ve gradually seen the Internet redefine many concepts in gaming, from worldwide multiplayer to purely digital distribution and publishing, things are only going to get more accessible and connected… eventually. For now, however, I wouldn’t fight too hard to be the first cautionary example of what doesn’t work yet.