Head in the Clouds

For a technology which is supposedly going to completely change the way we access and consume videogames, cloud gaming is certainly having a bumpy start. The latest chapter in the saga of OnLive, the hugely ambitious and much hyped cloud service, is perhaps the murkiest yet, with the company being shut down and its assets shuffled quickly into a new company, managed by the same CEO - leaving investors and staff in the original company high and dry.

If that sounds dodgy to you - as it does to a lot of people - that's because it is, albeit not in a legal sense. It's a piece of corporate manoeuvring which is permissible in the US, using a system called "Assignment for the Benefit of Creditors" (basically a pre-pack of a failing company, although the creditors in this instance are unlikely to feel that much has been done for their benefit), but while the legal ends may be neatly wrapped up, the morality of such a move - keeping the same people in charge while dumping staff and investors - is certainly deeply questionable.

"OnLive made some specific mistakes which reflect on extremely poor management and a deeply flawed understanding of the market."

Ours is not really the place to make moral judgments, though. What's more interesting is the question of what has come out in the wash regarding OnLive's market position - and what that says about the whole grand cloud gaming experiment.

The most oft-quoted figure, over the past week, is a number which purports to be OnLive's peak concurrent user figure - around 1800 users. It's tough to extrapolate subscriber numbers from that figure without additional data, but somewhere south of 20,000 seems plausible. Meanwhile, the service was costing around $5 million a month to run. This is not, in any sense, a viable business.

OnLive made some specific mistakes which reflect on extremely poor management and a deeply flawed understanding of the market. The company spent a fortune distributing OnLive streaming consoles for free in an extremely naive attempt to build an installed base - but few of those consoles are actually in use, not least because they were distributed to gamers who generally already own hardware capable of using OnLive services.

Many recipients also discovered that their broadband connections weren't up to OnLive's requirements; and of course, there's the simple reality that if you give someone something for free, they'll consider it to be of no value. A major part of the reason why the razors-and-razorblades model works for games consoles is because having made an investment in a console, users want to justify that investment by acquiring more games. With no investment in OnLive, users felt no such urge. Not only was OnLive wasting its money, it was actively devaluing its own service in users' eyes.

That speaks to the other fundamental flaw with the OnLive business. Who, exactly, was it targeted at? The marketing and promotion for the service seemed to be pretty directly targeted at core gamers and early adopters. Yet they're precisely the people who are most likely to be unwilling to suffer the flaws in OnLive's performance - the lag and graphical degradation which are an inevitable consequence of streaming high-definition interactive content over home broadband connections.

Moreover, OnLive's catalogue of software is largely made up of back catalogue titles, not the hot new releases which the core gamer market might be attracted to. This is a service which could potentially appeal to more casual players - people who don't keep up with the latest releases or want to buy the latest hardware, and don't mind a somewhat degraded experience in return for convenience. Yet it was pushed towards the early adopter market - not least, perhaps, because the casual player market probably doesn't have the broadband speed required for OnLive to work at all. Catch-22.

"It creates a business that's vastly expensive to run and difficult to scale, and which creates additional enormous step-costs every time you want to upgrade your basic hardware."

These are specific problems with OnLive's service, and there are a few more we could explore - but one could make the argument that they don't reflect a fundamental problem with the concept of a cloud gaming service in general. Indeed, there are plenty of cloud gaming apologists lining up to note that OnLive's failure is a consequence of problems with that company specifically, not with the concept behind it.

The reality is a bit tougher. If OnLive had succeeded, it would have answered a number of hard questions regarding cloud gaming - but by its failure, it leaves them hanging ominously over the head of any venture in this sector. The technological questions, regarding the quality of the experience, the speed of broadband, and so on, can be answered simply by the progress of time - broadband gets better, as does video compression technology, so it's undoubtable that we'll eventually be able to provide a cloud gaming service whose quality is indistinguishable from games running off locally executed code. The real question is, why would we want to?

That's not being facetious. My question about cloud gaming from the outset - a question which is only made more vital by OnLive's failure - is an economic one. Yes, it's possible (just about) to store game hardware in huge racks of servers at data centres rather than in boxes underneath people's TVs or in their pockets. But doing so transfers vast costs onto the service provider which have, up until now, been borne by the consumer in one way or another. It creates a business that's vastly expensive to run and difficult to scale, and which creates additional enormous step-costs every time you want to upgrade your basic hardware.

All the high-minded talk about freeing consumers from the need to buy or upgrade hardware looks slightly silly in an era when the most ubiquitous game devices are things like mobile phones and iPads - devices which are now possessed of extremely powerful 3D capabilities, and which consumers are probably going to buy or upgrade every couple of years anyway. The same is essentially true of computers; sure, your average laptop won't play Crysis 2 particularly well, but then again, nor will an OnLive style service - and your average laptop will certainly play the vast majority of games perfectly happily. Again, they're a device which consumers will pay to upgrade every few years anyway - so why, exactly, do cloud gaming proponents think it's a good idea for a games company to take on that expense, when consumers are clearly happy to bear it?

In fact, the only area in which this might possibly make any sense is in the field of game consoles - dedicated game hardware which, indeed, consumers might prefer not to have to buy or upgrade. However, OnLive can't actually play console games - and it's unlikely that any cloud service ever will, unless it's being offered by the console platform holder themselves.

In an era when processing, especially graphics processing, is getting vastly smaller, more efficient and cheaper - when the slim phone in your pocket will very soon pack more graphical punch than the console under your TV - we ought to be deeply suspicious of any business whose core proposition is to completely ignore the power of modern consumer hardware in favour of an outdated and unnecessary approach to centralised computation.

"The cloud, as a general concept, is going to be incredibly important for games."

Don't get me wrong - the cloud, as a general concept, is going to be incredibly important for games. Storing saves, user data and indeed the game data itself in the cloud is a great idea - one which many developers are already using (World of Warcraft, for example, grabs game assets from the cloud as you're playing, allowing players to get started with the game before the entire client has been downloaded).

But the notion of rendering those assets on a server and sending the frames to a dumb client is simply out of step with how modern technology works and what it's capable of. Yes, we can do it - but there's absolutely no convincing economic argument which says that we should do it, except in very limited usage cases such as game demos, which is why Gaikai (a significantly more cleverly focused and targeted company than OnLive) focused on that field rather than making a doomed attempt to launch a consumer service.

Does OnLive have a future? Not unless it can find a buyer, as Gaikai did. The company has some nice technology - but it does not have a viable consumer service. If, as some analysts have speculated, the existence of that consumer service has been actively putting off some potential buyers, then OnLive would do well to shut down its failed experiment and radically rethink what its business is for.