Very interesting article over at PC Perspective!  They give an overview of the PC version of OnLive and an in-depth assessment of the quality of the OnLive versions of the games.
 
It seems that at least for now fast paced games like Unreal Tournament 3 are indeed unplayable.  Also it seems that every title is running on significantly lower settings than standard.  The site has even put up videos of the games, and the performance and graphics look considerably worse than the console versions, let alone the PC versions at maxed settings.
 
Maybe things will clear up over time, but right now it looks like OnLive will be the worst way to play a game.

Guess there's still a lot of work to be done.

Well those people will be working on it for a while.

I'm sure OnLive could work with games that lag isn't that much of an issue (I'm not sure there are that many...), but yeah, it never seemed as if anything good was going to come from it.  It's supposed to be released later this year too, right?  That seems unlikely now.

Well it is a BETA so it is just to test stuff out and rework it all. You can't say that it representative of the final product.
 
 

I remain convinced that this will fail miserably. Their custom hardware is a neat idea, but the current infrastructure of the Internet is just not where it would need to be.
 
Plus, I can't imagine the servers they have actually running the games are that powerful.

what im dreading is the lag between input and reaction. having any delay will be bad for fast ass games. 

While all aspects might improve over time, and this is a REAL beta going on, the image compression looks absolutely atrocious right now (which I pointed out countless times before in prediction), but even I didn't expect THIS...
 

Well, I'm surprised is actually kinda works. 

No suprise there,Onlive is the next Gizmondo.

Well, we are getting close to april 1st again.

I love how it has the exact same problems everyone with a mind could put together through common sense. The idea is fantastic (or at least intriguing), but why are you trying so hard to implement something that simply isn't going to work with the foundations of the internet that we have laid out right now? I guess they're just trying to be first to market, but I imagine they'll find that market isn't there, soon enough.

Yep, it's not gonna work until the infrastructure can actually support it. Internet speeds are too poor right now, and they obviously don't have the games settings turned up which makes me think the hardware for hundreds of thousands of users isn't quite there yet either.

Compression artifacts and lag were a given with this service.  
 
The inherent problem is that the internet speeds are just not usable for the service to have any major value and I'd be surprised if it didn't stay valueless for at least another 3-5 years.  OnLive just isn't the same as services like Windows Update, iTunes, XBOX Live or PSN where big companies can leverage mirror services like Akamai to decentralise the data warehousing to more localised mirrors and so it will be highly unlikely that this service will succeed in my view.  Some are convinced that services like OnLive (if not the actual brand itself) will win out in the end, but I'm of the view that the investors will shy away from these kinds of offerings because the infrastructure and maintenance costs make services like these extremely top heavy investments with a lot of capitol at risk even before they go into testing, let alone once they've actually launched.  
 
Once launched the owners of such services have to bear the entirety of the system's maintenance costs which includes not only basic hardware, networking and operations software upkeep, but also the far more influential game selection and optimisations aspects.   Experience tells me that they are not ready for or indeed capable of managing such complex responsibilities and will more than likely be swamped with voluminous questions and complaints which they will not be able to cope with.  Gametap, an arguably simpler practice, was originally backed by Time-Warner/Turner Broadcasting but its countless revisions and maintenance headaches, including woeful 64bit compatibility, have since seen it change hands to be passed on (likely at a cheaper price then expected) to a similar french a service called  Metaboli who are now in the process of absorbing the Gametap library into their system.
 
Cable network management groups have seen the success of cable TV and potentially successful digital providers like Netflix and have just put two and two together hoping that the end result will be just as simple and easy to breed a gaming cash cow.  
 
They couldn't be more wrong.

Keep in mind that this was used with a beta account that was not his; he was outside of the geographical area of the beta so that may account for some of the lag issues. Although that doesn't make up for the graphics quality. 
 
I really don't think this will be even remotely possible on the scale and quality that we expect until 2015 or later. 

Oh man. That UT3 screenshot is atrocious.

" While all aspects might improve over time, and this is a REAL beta going on, the image compression looks absolutely atrocious right now (which I pointed out countless times before in prediction), but even I didn't expect THIS...
 

"

Well, it looks like the people at OnLive weren't happy about the truth getting out and in a petty attempt to mask that truth filed legal action against the preview site.  A DMCA notice required them to remove all the OnLive logos (yes, the logos, how pathetic) from their preview, which unfortunately includes the videos showing OnLive's poor performance.
 
The rest of the content (that which didn't have the OnLive logo anywhere on it) can stay online.

The Beta requires the participants to agree to a non-disclosre agreement, as well as certain tests to see if you meet the location, bandwidth, and performance mandates required to test, making the article grounds for legal action.  So it may appear that they're trying to cover up a negative showing, but it's because it was a preview of beta code on a non-agreed test platform, a legitimate reason.