What Went Wrong with OnLive?

Digital Foundry once said it couldn't possibly work, then grew to admire how close it came. Rich looks back at its history.

Two years after launch, OnLive is bankrupt and under new ownership: employees have been fired, their shares in the fledgling cloud outfit are worthless, and the original start-up has ceased to exist. What remains in terms of infrastructure, technology and IP is has been bought out by venture capitalist Gary Lauder, with OnLive 2.0 carrying on where its predecessor left off, re-hiring less than half of the original staff in the process. From a user perspective, it seems to be business as usual, with all servers remaining in service. However, in light of the facts and figures that have emerged over the weekend, it's difficult to imagine a rosy future ahead for the service - and for cloud gaming in general.

Various accounts of what actually happened last Friday at OnLive's Palo Alto HQ have made their way into the press over the last couple of days - with Joystiq's version of events the most detail-rich. However, the general facts and figures seem broadly consistent across all accounts: OnLive was burning through a staggering $5m per month, running 8,000 game servers that were only being utilised by 1,800 concurrent users - at peak. This, from what OnLive says is an active userbase of 1.5m gamers.

It's a remarkably small number for a service that's over two years old, having burned its way through so much investor capital. To be brutally frank, its concurrent user stats suggest a platform that is a niche irrelevance compared to its established rivals. To illustrate, that 1,800 number across the OnLive platform compares unfavourably with EVE Online's 50,000 peak in the last 24 hours and Team Fortress 2's 111,000 users in the same time period.

In retrospect, the 18-month-old $1.8bn valuation now looks patently ridiculous. OnLive claimed that it could revolutionise gaming, but the bottom line is that relatively speaking very few people bought into that. Literally.

Of course, with the benefit of hindsight, it's really not too difficult to see what went wrong with OnLive. As much as Steve Perlman would tell us that this was a viable replacement for the current generation of home consoles, by virtually every quantifiable criteria OnLive offers a sub-optimal means to play the exact same games you can already enjoy elsewhere. Visual quality is often impacted by dodgy video encoding and low bandwidth, and controller response is typically less than stellar compared to the established alternatives - and even its rival, Gaikai. OnLive claimed that it could offer a superior gameplay experience to the consoles, mentioning that it could run PC games at max settings, but the reality is that its datacentres used relatively meagre hardware by enthusiast standards, with the latest titles often running on very modest quality levels.

OnLive talked about console-beating experiences by running PC versions at max settings. In this OnLive/Gaikai Assassin's Creed: Revelations comparison, we could actually put this to the test. The Gaikai version has the graphics options available, allowing us to ramp them up to the highest level, showing a significant difference in quality. On the plus side, OnLive did manage to get very close to its target 60FPS... Use the full-screen button for full 720p resolution.

Content is King: The Real Reason OnLive Struggled?

Infrastructure issues continue to be a major issue for cloud gaming and consistently undermines the OnLive experience: the quality of service is entirely reliant on a decent internet connection with plenty of bandwidth, something which remains a luxury for most of the prospective userbase. Not only that, but to retain the best quality that connection cannot really be shared - even on a 50mbps fibre connection, gameplay suffered significantly when simultaneously uploading a file in our testing. On a standard 8mbs ADSL line, playing a YouTube video concurrent with OnLive gameplay could be disastrous to performance.

Back in 2009, rather famously we opined that OnLive Can't Possibly Work. "Realistically, there is no way it can work to the extent suggested, and no way it can provide a gaming experience as good as the one you already have without inherent compromises," we said.

In our first hands-on analysis out of controlled conditions (at events OnLive curiously manages to look better than any home connection we've experienced) the service mostly came into line with our initial expectations, but it was clear that there was real potential here, that we hadn't given full credit to the achievement. Indeed, in best case scenarios, even the "insurmountable" latency issue looked as though it could be resolved. But looking back over the past couple of years, we suspect it's not just the tech that caused OnLive problems - arguably, the situation with regards actual content was far more damaging.

This remains the key issue where OnLive fails to deliver. To this day, exclusives are non-existent and the entire catalogue can be played on PC, with all major AAA titles also available on console. Many titles only appear on OnLive much later after they appear elsewhere. Perhaps more importantly, for a system designed for accessibility and best suited to attract the non-hardcore, the vast majority of OnLive's catalogue is aimed squarely at the committed gamer. The cost is also an issue too - boxed game prices are subject to competitive market forces that bring prices down. OnLive's fixed high prices can end up being a more expensive way to play a sub-optimal version of the same game you can already play elsewhere. Consumer concerns about losing access to their library of titles is also clearly an issue in the light of what happened last week: OnLive's new owner needs to make an unambiguous statement of intent on his future plans for his new acquisition.

"OnLive never lived up to the grandiose claims of its management, but despite its many shortcomings there's no denying it's a remarkable, pioneering first-gen product."

Deus Ex: Human Revolution worked fairly well on OnLive, beating console performance, though frame-rates varied dramatically from a console-beating 60FPS in less taxing areas to half that in the hub cities. However, graphical presets seemed to be set to a significantly lower level than console - not quite the 'max settings' OnLive users might have hoped for.

A Pioneering Service Ahead Of Its Time?

On the flipside, despite its many issues, OnLive remains an enormous achievement in many ways, with plenty of attractive features. The basic fact that it works at all is a miracle. While best described as first-gen tech with many deficiencies, OnLive at least sought to address the underlying problems facing cloud gaming. And when it works well, it can be a remarkable experience: there is undoubtedly something rather magical about seeing it in action, especially when running a game better suited to the limitations of the video compression technology. Play OnLive on a smaller screen with a relatively high pixel density - like the Xperia Play smartphone for example - and the effect can be extraordinary, as the macroblocking artifacts are much less noticeable. While implementation was inconsistent to say the least, OnLive also targets 60 frames per second - something that consoles and competing service Gaikai can't match.

OnLive breaks new ground in other areas too: every gameplay instance is beamed to a central server, meaning that users can spectate on the action via its Arena function - something that would be next to impossible to implement on other online services. The vast majority of the catalogue has free demos too, allowing for 30 minutes of gameplay on the final game before asking for any money. While its new game pricing remains fundamentally flawed, the "all you can eat" PlayPack, offering access to hundreds of games for an attractive monthly flat cost, remains an excellent idea.

In terms of features, OnLive is competitive: voice chat is there now, achievements are implemented and online multiplayer works. Other fundamental elements of the OnLive proposition remain as attractive as they always have been - the ability to run gameplay on any device that can decode 720p60 h.264 video, for example, along with the superbly designed micro-console. OnLive even got it right with the controller too - something that isn't as easy as it may sound.

But, as the new owners of OnLive take over, it's safe to say that the immediate future of cloud gaming is looking grim. The relatively miniscule userbase points to an inescapable conclusion: there is something very wrong with the basic proposition as it stands in the here and now.

"The relatively miniscule OnLive userbase points to an inescapable conclusion: there is something very wrong with the basic proposition as it stands in the here and now."

We used to think that latency in cloud gaming could never be adequately tackled. But in best-case scenarios, OnLive worked out pretty well, and in this video we see Bulletstorm on Gaikai matching the Xbox 360 version's input latency. However, generally speaking, the consistency just isn't there - cloud needs a generational leap in broadband quality to compete with the local experience on fast-action games.

For cloud gaming to truly become a contender, there are still significant issues to resolve in terms of latency - though technologies such as GeForce GRID are clearly a step in the right direction. For streaming gameplay to overcome the visual inconsistencies and reliance on an unshared connection, a generational leap in the bandwidth and quality of broadband internet is required to make it truly competitive.

For OnLive, the unfortunate truth of the matter is that nobody really knows when super-speed internet will roll out to enough people to allow this to happen. Three years perhaps? Five? Ten? Regardless, progress is likely to be gradual, and it's really difficult to imagine OnLive remaining sustainable across the duration. The genie is out of the bottle: now the userbase figures are available, who is going to put any effort into porting their games to this most niche of platforms?

Perhaps more important than the tech is the content and this is where Sony's acquisition of Gaikai could make the difference. Even in its current state, with all the limitations of current-gen broadband, the cloud is a platform just like any other - it has fundamental strengths and weaknesses. If developers target the plus points of the technology and steer around the negative aspects, we could see new games being developed that simply wouldn't be possible on existing hardware.

Imagine a game that plays to the strengths of the specific server hardware, where developers can tap into the video streams of every player and expand game worlds and content without ever having to roll out client-side patches. Just the potential of server-side access to an immense internet connection offers up a huge amount of potential. In the right hands, the possibilities are breathtaking. Sony is best placed to address the exclusive content issue - it has a phenomenal wealth of in-house talent that could make the most of the platform. Why bother streaming existing titles that may not play as well when you can create brand new games that run on all existing Sony current and next-gen consoles, tablets and smartphones? What kind of game could Naughty Dog or Sony Santa Monica build from the ground up with cloud hardware in mind? It's a mouth-watering proposition.

But where does this leave OnLive, with no in-house game development talent? What we have is a digital distribution platform offering third-party games content in no way tailored to the potential strengths of the underlying technology, and a tiny userbase that rules out AAA cloud-optimised titles from third-party publishers. Steve Perlman's outfit has finally come face to face with reality and it will be fascinating to see where the company goes next.

Comments (64)

Comments for this article are now closed, but please feel free to continue chatting on the forum!