If the network latency <150ms/and speed is enough to stream its fine. The cost of processing video vs dynamically generating everything in the game is significantly less.
Won't be usable, too much latency between when input is created and when the video feed is updated. On my ISP, there's a 35ms ping just to the edge of my ISP's routers. Even if the onlive system provides no delays on either end, that's way too long for a realtime game. To tell you the truth, I don't think that would be usable even on a turn based game. And god help anyone running over a wireless connection.
Of course, gaming media doesn't seem able to wrap their heads around the difference between a controlled LAN setup onlive was demoed on and an actual customer internet connection.
I wrote a big explanation but I'll make it short for you:
tl;dr: No fucking way, it's just another WEB 2.0 sham. They're just getting as much investor money as they can. Even if it was technically feasible, which right not it isn't, it wouldn't make economical sense.
hey guys, let's get people to pay to play video games over vnc!
Name:
Anonymous2009-03-24 22:25
Yeah, no way this could be usable when games with local physics are near-unplayable with high ping. 150ms for your input to hit the server and another 150 for the video to get back to you is not an unreasonable estimate, especially in shitty connection land where it could be even higher. It is, however, unreasonable to think players are going to play with lag approaching a half second. Even setting aside the bandwidth issues and the fact that they would need the rendering capacity to render these games for every person playing (how many players can we even render on one server?), this is nuts.