Why your next game console ought to be Watson

A few years ago, the gaming world was thrilled by the premise that the cloud (the Cloud!) could be harnessed to power games, too — any game you wanted, anytime, on any device, served from data centers to you. Services like OnLive and Gaikai promised freedom from your hardware, the end of the lockout of exclusive games only available on one platform or another.

Reality disappointed: What we actually got was a limited library of not-new games (Homefront, anyone?), many of which you already owned, but even laggier than on your own hardware. Turns out traditional retail, game publisher, and hardware platform companies made it difficult for cloud gaming services to get the best games on the day of release, and even then the gameplay quality was slightly inferior.

But the concept of gaming in the cloud is still an idea worth pursing for a far greater promise: the ability to deliver an entirely new kind of game experience.

Historically, in games as in any other media, new distribution technologies enable new creative experiences. Pong wouldn’t have been possible without a new device plugged in to your TV. Internet-connected computers meant you could play Duke Nukem and Quake with other people online. The evolution of server technology brought massively-multiplayer games. The iPhone brought Angry Birds, a game designed for a touch interface, and so forth.

So why should a cloud gaming service be used to deliver the same old games as before that were built for a $250 machine?

What we should be wondering, then, is what new kinds of games and gaming experiences cloud delivery could inspire? Compared to the gaming hardware you own, a cloud gaming service could access much more computing power—with a limitless capacity to add processing. Consider after all that the most powerful supercomputer in the world, the Titan, is about 70,000 times more powerful than an Xbox 360. Granted the Titan costs a cool $100 million, which cuts out most households, but scaling back to basic and accessible data center prices would still offer many orders of magnitude more computing power than any current or near-future home console . (And this isn’t to say great gaming experiences are limited to powerful hardware—to the contrary mobile phones play compelling games, too. They’re just of a different sort.)

As for content itself, games purpose-built for the cloud do not yet exist — ones that aren’t encumbered by the limits of processing power, that would use the full advantage of many more, and more powerful, CPUs and GPUs. These “supercomputer games” would open up creative possibilities far beyond what games of today are capable.

Imagine supercomputer games with vividly lifelike worlds and characters (and not the almost-real, uncanny valley of current-generation graphics), or a single battlefield with 50,000 other players playing at the same time — or opponent AI on the level of IBM’s Jeopardy!-winning Watson. Supercomputer games could be dramatically different from anything you can play tonight at home. I’m no game designer, but what if we could use real-time traffic data to fill the streets of the next Grand Theft Auto, or step into a computer-generated world that looks as compelling as the Lord of the Rings movies?

Now, there are many reasons, beyond the technological, that these games don’t yet exist: It would be prohibitively expensive to pay artists to create all those detailed graphics, and simple AI is good enough to defeat most any player at most any game. But the record of creative innovators is that eventually they find a way to stretch the available technology to its limit. And some gamemakers are already beginning to probe at the games you can create if you host some of the game in the cloud.

There is a nagging constraint to the cloud, of course — bandwidth, which simply isn’t growing at the pace of Moore’s law. Network latency makes fast-twitch games, in which defeat is determined in microseconds (like with the top console genre, first-person shooters) hard to play over today’s internet. So, at least until the next engineering breakthrough, these supercomputer games might be designed around genres requiring slower player reflexes than, say, Call of Duty or StarCraft.

Best of all, the only hardware you would need at home is a basic input device like a controller and a box to render the graphics, and it could be cross-platform so that you could play from a PC or Mac or any smartphone. As one for-instance, OUYA, the new open, Android-based console I back, could be great for a cloud-delivered game (hear me, developers?), and its notable that Sony bought up Gaikai and certainly has plans.  (Full disclosure:  OUYA also has an announced partnership with the relaunched OnLive.)

Supercomputer games could be extraordinary. Now some intrepid game developers just have to make one.

Roy Bahat is Chairman of the open, Android-based game console company OUYA, and is former president of IGN. He is also on the faculty at UC Berkeley. Follow him on Twitter @roybahat