Since folks continue to ask questions about OnLive, I thought I’d follow up a bit on my earlier post, and include the information I gathered from discussions with OnLive at their GDC booth.
I liked the guys I spoke with, and they seemed open and forthright about the product, giving me the feeling that OpenLive isn’t bunk so much as the product of good technical people creating something moderately useful. But that product has, for strategic purposes, been positioned by their marketing and biz dev folks as something it truly is not; competitive with existing products or in any significant way market changing.
When I quizzed these guys on how well the system would work in the real world, they were open about its limitations. The Internet is squirrely, and the way that most folks configure home networks (wifi, multiple users, etc.) degrades performance, while an instantly streamed video needs ideal performance. You really couldn’t play a frontline game on OnLive. The most you could do is get a basic feel for the graphics and gameplay, so it could usefully be positioned as a marketing device for frontline games, and I can see this being a worthwhile use of the technology. -The easy gut check on this is to think of clicking on any content on a webbrowser, and how often you have to wait a moment for connection and download to start, even with an edge network employed.
They did advocate for OnLive being a good place to play less demanding or casual games, but I don’t get how that business needs this sort of technology. Casual has a fully functional model in browser-based and small executable games, which tend to require similarly minimal processing power, so OnLive’s posited server side GPU advantage is basically obviated. When I first saw this sort of tech demo-ed by g-cluster for Comcast some years ago, it was intended to leverage the broadband network to help minimize the barrier to gameplay of cruddy settop boxes from MSOs. There’s a lot more going on around the television these days, so that’s less important, especially for anyone with an interest in games, even casual. My own feeling, though, is that the big push by OnLive to present itself as significant and strategic is largely to achieve presence at the headend of broadband providers (especially MSOs), because I don’t believe it will work well enough for any sort of games unless they have ubiquitous presence at the headend. -This is where the questionable feel of the thing comes in; they need to sound game-changing, but they’re not, and anybody that’s familiar with game and/or internet tech knows that they’re not.
Where the guys at OnLive almost had me on their product having a unique value was in their description of video compression and decompression technology. My feeling generally is that the problem models like OnLive must deal with best is optimizing video streaming, and that with all of the major entities putting significant resources to this problem, I don’t see how a game-video-streaming entity can compete. Hence, their core technology is never defensible.
But the fellow at OnLive responded that games have a unique need for immediate “bufferless” video with instant compression and decompression. Which seemed valid at the time, but not completely sayisfying to me. Upon further consideration, my opinion is this; while it is true that games have a unique need, all that “instant” compression and decompression actually is is highly efficient compression and decompression. And highly efficient compression and decompression is the holy grail for a number of powerful and important entities. The game-specific focus doesn’t really drive the need for a different sort of compression, it just changes the requirements for compression of the stream, most likely forcing the 1:1 ratio of “servers” to users that OnLive utilizes. The unique aspect of video for games distributed in this manner is in making the video feel like an unbroken experience, rather than sequential pieces, somehow masking the ugliness of connection irregularities, but that’s likely just a matter of aesthetics.