Why I think OnLive will fail
OnLive has been making a big stir at GDC2009, but I think it will be a flop. The idea behind OnLive is being able to play your game whereever you are, with whatever devices you have, even if they’re not powerful enough to run said game (Crysis on a netbook would be a good example of this). OnLive achieves this by running the games on their servers ‘in the cloud’, and streaming it to you as a video.
Why it will fail in Australia/other countries
- Bandwidth
For 800×600, 1.5Mbit/second (1500kbit/second) or 187KB/s is needed. For 1280×720 ("720p"),4.5Mbit/second or 562.6KB/s is neededturns out its 5MBit/s or 640KB/s. - Quality
As mentioned above, a fairly insane amount of is needed to go to 720p. My monitor runs at 1680×1050 (it’s a 20" but 22" monitors also run at this resolution) – just under twice the amount of pixels. Stretching the image to almost twice its size isn’t going to look good, particularly when you factor in any ‘blocks’ generated by the encoding process.
If it has the option to choose your resolution, my monitors native resolution has twice as many pixels as the 720p option, meaning I’d need 1MB/s to play! - Data allowance
Australia is known for slower speeds and smaller download quotas, and bandwidth intensive applications can be rather expensive For an hour of gameplay in 720p, I’d churn through just under two gigabytes*. If I was to play an hour a week, that’d be around 8GB/month. What if I’m not the only gamer in the house (and I’m not)? 16GB! On Telstra (not the cheapest, but biggest), that’d set you back a minimum of INSERT NUMBER (by comparison, currently playing a game online usually doesn’t consume more than 100MB/hour, or 800MB/month in this usage scenario)* 562.6 (rate per second) * 60 (per minute) * 60 (per hour) / 1024 (making it megabytes instead of kilobytes )= 1977MB.
- Latency
The biggest problem with gaming with international people is the lag, which is why there are lots of Australian based servers. When the game itself is being served from the US, that’s about a minimum of 250ms delay (make that 500ms – 250ms for you to send a keystroke/mouse movement, 250ms for the results to be sent back to you)The obvious solution is to have local (at least "continent" level) servers. The downside to this is that it would likely drive up the subscription cost (I presume there will be a subscription to the OnLive service) to regions other than US.
- Cost
At the end of the day, its going to cost them money to be running the games, and that is going to hit users of the service in the form of subscription fees. Too low isn’t going to make them successful, too high and it’d really be a debate as to if its worth it when you consider the other problems with the server (mentioned above).
There has been no information released about price
So if it’s going to fail, why are publishers on board?
The biggest defence of the OnLive system has been "this is the next big thing because big name publishers are supporting it’. The market for this type of system is people who can’t afford (or refuse) to buy a computer capable of playing all the games they want. From a publishers perspective, even if OnLive goes bust after a month of operation, that is a market they wouldn’t have tapped even if they gave the game away for free.
My feelings are if it didn’t have the support of at least one major publisher, it wouldn’t even get past the ‘this is a cool concept’ stage, and would end up being like the Phantom console.
Good idea, wrong era
If we fast forward ten, twenty, maybe thirty years time, when everybody has a minimum of a single fibre connection to their house, all the limitations will disappear, and then ‘dumb terminals’ doing all the processing on ‘the cloud’ will be the way to go.