Mario meets cloud computing.
There’s more to digital distribution than just online sales – being able to instantly buy and (sometimes) play games online is only a part of the puzzle. Where the game runs, where the game is stored, and how you “own” the game are equally important. Steam, Good Old Games, and similar services are the simplest model; by and large, they replicate bricks and mortar stores by providing a catalogue and a distribution system (in their case, the internet). As a category, these “Games on Demand” system are most likely to be the dominant model for the foreseeable future.
That’s not to say they’re the only model. The badly termed “games as a service”, for example, abstracts the whole concept of property and hardware requirements and shifts all processing and storage server-side. In effect, the gamer either licenses or rents a game and receives nothing more than an A/V-stream – their savegames, their purchases, and even their computation lives in the cloud. OnLive has been the most public in this space.
It’s a neat idea – instead of only having one real opportunity to upgrade your console per generation or having to spend significant amounts of money upgrading your PC, the client-side device only needs to be able to consistently stream and decode video at an acceptable resolution. Generational hardware upgrades could then be incremental and capitalise on economics of shared infrastructure, reducing marginal hardware costs by exploiting the fact that not all gamers play games at the same time. Similar to sharing internet capacity, the provider needs only use a single “server” across multiple gamers. Plus, the provider can offer a theoretically infinite catalogue with no install time required – everything’s instantly available in the cloud. All you need is an internet connection with sufficient bandwidth and a (relatively) dumb box to handle the streaming and provide gamepad connectivity. Cool, hey?
While it’s entirely possible these streaming services may actually get some traction and provide a credible alternative, I have my doubts. Here’s why, in a nutshell:
- Latency
- Cost
- Pricing
Latency
Streaming a graphically intensive game is tough, very tough. Latency isn’t really an issue in games like Bejewelled or Dragon Age: Origins, but in action-heavy games, it can be a killer; most bands struggle to play well together if there’s more than a 30 millisecond delay between instruments, and 80-100 milliseconds appears to be the absolute upper limit of what’s acceptable to a twitch gamer. On top of that, there’s now a general expectation that the minimum “standard” for full-featured gaming is a minimum 720p resolution (the Wii notwithstanding) and we’re rapidly approaching the point where 60fps is an expected feature.
Ignoring the bandwidth requirements, in that 80-100 milliseconds any streaming service will have to go through the following processing chain: Player Action -> Transmission -> Game Logic (AI, Sound, etc) -> Image Rendering -> Image / Sound Compression -> Transmission -> Image / Sound Decompression -> Image Display. That’s a lot to fit into a single time-constrained path given the vagaries of operating over a public packet-based network. Get it wrong, and you’ve got some very unhappy gamers.
Cost
Within this service, effective delivery means minimising the individual time components within the overall processing chain. Dedicated hardware and smart coding will help with some of these, but the transmission path is a key bottleneck – not only is the routing outside of the control of systems operator, but it needs to be traversed not once but twice! It’s all well and good to minimise the total path by locating the hosting environment closer to the gamer, but that’s not cheap – fundamentally, that means maintaining a dedicated data centre in potentially every city where the service is offered.
While the labour costs alone make that a highly debatable proposition, what happens when there isn’t critical mass to warrant setting up a datacentre in a given city? Given the staggering lack of Australian-based MMO server farms, the odds of any such service actually making it outside of London, New York, Los Angeles, and Tokyo (where population figures suggest a large enough market to make such a service sustainable) are slim at best.
Pricing
Similar to digital print, most people’s expectations about likely discount levels for digitally distributed goods rather than physical goods is vastly overestimated. It’s an understandable assumption – they’re digital, so there’s no physical cost beyond storage. Bandwidth is cheap compared to air distribution and physical manufacturing, and while there are diseconomies of scale associated with storage costs, games catalogues aren’t anywhere near the point where serious diseconomies kick in yet.
History’s shown otherwise, unfortunately. When online book stores launched, many had an expectation that they’d be buying all their $19.95 books for under $5. As time has shown, that’s just not the case, and there’s good reason for it; the entire industry is built on a certain level of return.
With games, the retailer’s margin on a typical $80 game sale in Australia is easily under AU$10 – RRP may be upwards of AU$110, but I’m betting that few games actually sell for anything near RRP. Physical manufacturing and distribution costs may add in another $5-$10 or so when amortised across the full shipment. So, out of that AU$80 software sale, there’s only AU$20 of “slack”, AU$10 of which the distributor will probably be wanting to re-appropriate for themselves.
Modern games are expensive to make, and developers and publishers have a certain level of profit expectations. Streamed games may have the potential to offer a cheaper alternative to physically distributed goods, but the reality is that the final retail pricing differential is likely to be marginal (if it exists at all).
So where does that leave us? It’s possible that such a service may actually get off the ground (anything’s possible), but the odds are against it. Assuming the technical hurdles are properly surmounted (a big assumption) and assuming there’s sufficient demand to make it a global offering (an even bigger assumption), it’s still caught in the same trap as general digital distribution – the value’s in the catalogue. And, getting a decent catalogue is hard in its own right.
These aren’t trivial challenges – much like the Phantom, I’m betting that interest in services like OnLive will be equally as ephemeral.
It does seem unlikely that we’ll see blockbuster action games in the vein of Modern Warfare streamed from a server. On the other hand, smaller, slower-paced games could be feasible. For example, World of Goo could be streamed without latency problems, since the graphics require less bandwidth and the game demands less instantaneous responses from the player. That would necessitate fewer server farms, which keeps the price down along with the relatively cheap development budget. It’s not a long jump from what’s playable from a web browser today, for free.
I think you’re right about high definition action games, and probably even slower-paced high definition games like Dragon Age, but I also think we could see something like a Pokémon game streaming before long.
The lack of a physical copy of the game will be a concern, though. What happens if the server farm burns down? Or simply if the distributor goes out of business? The psychological satisfaction of possessing a game shouldn’t be underestimated either; if people don’t get the same consumerist thrill from acquiring something new and shiny, the economically rational decision might just not sway them.
I think you’ve hit it – ironically, I think the biggest opportunity is actually in the expanded market, not in the core market. I don’t think OnLive understands this; effectively, they’re competing against Sony and Microsoft. I can only assume that they’re hoping they’ll be bought out, as there’s no way they’ll be able to compete. I mean, Sony and Microsoft haven’t even made a return yet this generation!
Here’s my offbeat prediction for the week – OnLive is going to get bought by someone like PopCap. Don’t think of it as a console, think of it as an alternative (and cheap) distribution channel. Because the processing requirements are low, the incremental cost of computation in the serverfarm is also low. Latency isn’t a huge issue so they can create a more regionalised approach with their datacentres. And, if they use a subscription model in conjunction with someone like Big Fish (who release a game every day), they’ve got a watertight value proposition.
Unlike others, I’m not so worried about the lack of physical goods – the number of collectors are very small. Think how many people subscribe to cable and you’re onto a good indication about how much the majority of people care about “owning” their digital entertainment. It’s a different model and lends itself more to subscription, but it’s the future for the extended market (even if it’s simply more of a leasing or check-in / check-out model).